PROGRAMMING

YOLO (apple, orange / 교육) 본문

OpenCV/머신비전 - 이미지 디텍팅

YOLO (apple, orange / 교육)

Raccoon2125 2020. 12. 24. 16:46
FruitCustomYolo

1. Mount Google Drive

In [1]:
from google.colab import drive
drive.mount('/content/drive')

Mounted at /content/drive
In [8]:
!ls -la "/content/drive/MyDrive/darknet"

total 24
drwx------ 2 root root 4096 Dec 22 21:36 bin
drwx------ 2 root root 4096 Dec 22 23:01 cfg
drwx------ 2 root root 4096 Dec 22 21:38 cuDNN
drwx------ 3 root root 4096 Dec 22 23:01 data
drwx------ 2 root root 4096 Dec 22 21:36 .ipynb_checkpoints
drwx------ 2 root root 4096 Dec 22 21:44 weights
In [9]:
!ls -la /usr/local/cuda/include/cudnn.h

-r--r--r-- 1 root root 134857 Sep 26  2019 /usr/local/cuda/include/cudnn.h

2. Install CUDA related modules, cuDNN

Unzip the cuDNN from Google Drive to Colab /usr/local/cuda folder

In [10]:
!tar -xzvf /content/drive/MyDrive/darknet/cuDNN/cudnn-10.1-linux-x64-v7.6.5.32.tgz -C /usr/local/
!chmod a+r /usr/local/cuda/include/cudnn.h

!cat /usr/local/cuda/include/cudnn.h | grep CUDNN_MAJOR -A 2

cuda/include/cudnn.h
cuda/NVIDIA_SLA_cuDNN_Support.txt
cuda/lib64/libcudnn.so
cuda/lib64/libcudnn.so.7
cuda/lib64/libcudnn.so.7.6.5
cuda/lib64/libcudnn_static.a
#define CUDNN_MAJOR 7
#define CUDNN_MINOR 6
#define CUDNN_PATCHLEVEL 5
--
#define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)

#include "driver_types.h"

3. Load DarkNet

In [4]:
import os
if not os.path.exists('darknet'):
    os.makedirs('darknet')
%cd darknet
%ls

/content/darknet
In [11]:
!ls -la '/content/drive/MyDrive/darknet/bin/darknet'

!cp /content/drive/MyDrive/darknet/bin/darknet ./darknet

!chmod +x ./darknet

-rw------- 1 root root 3213624 Dec 22 21:37 /content/drive/MyDrive/darknet/bin/darknet

4. Test Yolo

In [12]:
!cp -r '/content/drive/MyDrive/darknet/weights' .
!cp -r '/content/drive/MyDrive/darknet/cfg' .
!cp -ar '/content/drive/MyDrive/darknet/data' .
%ls

cfg/  darknet*  data/  weights/
In [13]:
def imShow(path):
  import cv2
  import matplotlib.pyplot as plt
  %matplotlib inline

  image = cv2.imread(path)
  height, width = image.shape[:2]
  resized_image = cv2.resize(image,(3*width, 3*height), interpolation = cv2.INTER_CUBIC)

  fig = plt.gcf()
  fig.set_size_inches(18, 10)
  plt.axis("off")
  #plt.rcParams['figure.figsize'] = [10, 5]
  plt.imshow(cv2.cvtColor(resized_image, cv2.COLOR_BGR2RGB))
  plt.show()
  
  
def upload():
  from google.colab import files
  uploaded = files.upload() 
  for name, data in uploaded.items():
    with open(name, 'wb') as f:
      f.write(data)
      print ('saved file', name)

      
def download(path):
  from google.colab import files
  files.download(path)
In [14]:
!./darknet detect cfg/yolov3.cfg weights/yolov3.weights data/fruit10.jpg
imShow('predictions.jpg')

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv    255  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 255 0.088 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv    255  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 255 0.177 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv    255  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 255 0.353 BF
 106 yolo
Total BFLOPS 65.864 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from weights/yolov3.weights...
 seen 64 
Done!
data/fruit10.jpg: Predicted in 41.828000 milli-seconds.
orange: 99%
orange: 59%
apple: 85%
orange: 99%
Unable to init server: Could not connect: Connection refused

(predictions:450): Gtk-WARNING **: 23:08:49.618: cannot open display:

5. Custom Yolo

In [17]:
!cp -r "/content/drive/MyDrive/darknet/custom" .

아래를 누르면 아주 긴 시간 동안 학습하게 됨

In [ ]:
!./darknet detector train custom/custom_data.data custom/custom-train-yolo.cfg weights/darknet53.conv.74 -dont_show 

custom-train-yolo
layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from weights/darknet53.conv.74...
 seen 64 
Done!

 Warning: You set batch=32 lower than 64! It is recommended to set batch=64 subdivision=64 
Learning Rate: 0.001, Momentum: 0.9, Decay: 0.0005
Resizing
608 x 608 
 try to allocate additional workspace_size = 108.69 MB 
 CUDA allocate done! 
Loaded: 1.363494 seconds

 1: 1874.486450, 1874.486450 avg loss, 0.000000 rate, 4.645283 seconds, 32 images
Loaded: 0.167962 seconds

 2: 1873.601929, 1874.397949 avg loss, 0.000000 rate, 4.587090 seconds, 64 images
Loaded: 0.172551 seconds

 3: 1872.306152, 1874.188721 avg loss, 0.000000 rate, 4.604361 seconds, 96 images
Loaded: 0.103026 seconds

 4: 1873.848022, 1874.154663 avg loss, 0.000000 rate, 4.703347 seconds, 128 images
Loaded: 0.029995 seconds

 5: 1873.857788, 1874.125000 avg loss, 0.000000 rate, 4.806937 seconds, 160 images
Loaded: 0.000041 seconds

 6: 1872.471313, 1873.959595 avg loss, 0.000000 rate, 4.846819 seconds, 192 images
Loaded: 0.000034 seconds

 7: 1872.012573, 1873.764893 avg loss, 0.000000 rate, 4.917636 seconds, 224 images
Loaded: 0.000040 seconds

 8: 1871.453247, 1873.533691 avg loss, 0.000000 rate, 4.951915 seconds, 256 images
Loaded: 0.000040 seconds

 9: 1873.198608, 1873.500244 avg loss, 0.000000 rate, 4.973237 seconds, 288 images
Loaded: 0.000042 seconds

 10: 1874.567993, 1873.607056 avg loss, 0.000000 rate, 5.029345 seconds, 320 images
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 3.362629 seconds

 11: 1500.653931, 1836.311768 avg loss, 0.000000 rate, 4.252978 seconds, 352 images
Loaded: 0.282405 seconds

 12: 1501.095581, 1802.790161 avg loss, 0.000000 rate, 4.167760 seconds, 384 images
Loaded: 0.349002 seconds

 13: 1499.952393, 1772.506348 avg loss, 0.000000 rate, 4.227568 seconds, 416 images
Loaded: 0.362561 seconds

 14: 1499.190674, 1745.174805 avg loss, 0.000000 rate, 4.260640 seconds, 448 images
Loaded: 0.304504 seconds

 15: 1499.838745, 1720.641235 avg loss, 0.000000 rate, 4.286100 seconds, 480 images
Loaded: 0.270931 seconds

 16: 1499.627319, 1698.539795 avg loss, 0.000000 rate, 4.326952 seconds, 512 images
Loaded: 0.258509 seconds

 17: 1498.648438, 1678.550659 avg loss, 0.000000 rate, 4.361765 seconds, 544 images
Loaded: 0.213899 seconds

 18: 1498.489380, 1660.544556 avg loss, 0.000000 rate, 4.400581 seconds, 576 images
Loaded: 0.138625 seconds

 19: 1499.743286, 1644.464478 avg loss, 0.000000 rate, 4.442517 seconds, 608 images
Loaded: 0.087263 seconds

 20: 1499.915894, 1630.009644 avg loss, 0.000000 rate, 4.484985 seconds, 640 images
Resizing
352 x 352 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 2.695915 seconds

 21: 631.258118, 1530.134521 avg loss, 0.000000 rate, 2.010143 seconds, 672 images
Loaded: 2.083407 seconds

 22: 629.917053, 1440.112793 avg loss, 0.000000 rate, 1.985406 seconds, 704 images
Loaded: 2.104342 seconds

 23: 630.067505, 1359.108276 avg loss, 0.000000 rate, 1.997155 seconds, 736 images
Loaded: 2.034616 seconds

 24: 629.935181, 1286.190918 avg loss, 0.000000 rate, 1.989491 seconds, 768 images
Loaded: 2.002633 seconds

 25: 629.285522, 1220.500366 avg loss, 0.000000 rate, 1.951859 seconds, 800 images
Loaded: 2.065752 seconds

 26: 630.179565, 1161.468262 avg loss, 0.000000 rate, 1.977708 seconds, 832 images
Loaded: 2.153957 seconds

 27: 629.511108, 1108.272583 avg loss, 0.000000 rate, 1.991136 seconds, 864 images
Loaded: 2.068506 seconds

 28: 628.575745, 1060.302856 avg loss, 0.000000 rate, 1.955530 seconds, 896 images
Loaded: 2.115319 seconds

 29: 628.781921, 1017.150757 avg loss, 0.000000 rate, 1.964389 seconds, 928 images
Loaded: 2.043752 seconds

 30: 629.131775, 978.348877 avg loss, 0.000000 rate, 1.989250 seconds, 960 images
Resizing
352 x 352 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.562863 seconds

 31: 629.188904, 943.432861 avg loss, 0.000000 rate, 2.005354 seconds, 992 images
Loaded: 1.989636 seconds

 32: 628.645874, 911.954163 avg loss, 0.000000 rate, 1.981603 seconds, 1024 images
Loaded: 2.082400 seconds

 33: 629.026672, 883.661438 avg loss, 0.000000 rate, 2.014967 seconds, 1056 images
Loaded: 2.098704 seconds

 34: 628.673645, 858.162659 avg loss, 0.000000 rate, 1.999370 seconds, 1088 images
Loaded: 2.032523 seconds

 35: 628.562927, 835.202698 avg loss, 0.000000 rate, 2.004468 seconds, 1120 images
Loaded: 1.980003 seconds

 36: 627.727600, 814.455200 avg loss, 0.000000 rate, 2.038285 seconds, 1152 images
Loaded: 1.991357 seconds

 37: 628.312500, 795.840942 avg loss, 0.000000 rate, 2.043637 seconds, 1184 images
Loaded: 1.977712 seconds

 38: 627.538879, 779.010742 avg loss, 0.000000 rate, 2.025044 seconds, 1216 images
Loaded: 2.026318 seconds

 39: 627.417847, 763.851440 avg loss, 0.000000 rate, 2.020436 seconds, 1248 images
Loaded: 2.010043 seconds

 40: 626.086731, 750.074951 avg loss, 0.000000 rate, 1.994252 seconds, 1280 images
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 1.685651 seconds

 41: 1489.795166, 824.046997 avg loss, 0.000000 rate, 4.347408 seconds, 1312 images
Loaded: 0.302650 seconds

 42: 1487.847778, 890.427063 avg loss, 0.000000 rate, 4.293712 seconds, 1344 images
Loaded: 0.251668 seconds

 43: 1486.417114, 950.026062 avg loss, 0.000000 rate, 4.301047 seconds, 1376 images
Loaded: 0.184303 seconds

 44: 1485.891357, 1003.612610 avg loss, 0.000000 rate, 4.339316 seconds, 1408 images
Loaded: 0.200205 seconds

 45: 1483.928589, 1051.644165 avg loss, 0.000000 rate, 4.351284 seconds, 1440 images
Loaded: 0.150421 seconds

 46: 1481.571655, 1094.636963 avg loss, 0.000000 rate, 4.350120 seconds, 1472 images
Loaded: 0.134612 seconds

 47: 1477.576416, 1132.930908 avg loss, 0.000000 rate, 4.392931 seconds, 1504 images
Loaded: 0.163153 seconds

 48: 1477.783936, 1167.416260 avg loss, 0.000000 rate, 4.324373 seconds, 1536 images
Loaded: 0.217562 seconds

 49: 1473.206665, 1197.995361 avg loss, 0.000000 rate, 4.303249 seconds, 1568 images
Loaded: 0.221576 seconds

 50: 1468.955444, 1225.091431 avg loss, 0.000000 rate, 4.293281 seconds, 1600 images
Resizing
608 x 608 
 try to allocate additional workspace_size = 108.69 MB 
 CUDA allocate done! 
Loaded: 0.669007 seconds

 51: 1829.339844, 1285.516235 avg loss, 0.000000 rate, 5.039371 seconds, 1632 images
Loaded: 0.000043 seconds

 52: 1823.746826, 1339.339355 avg loss, 0.000000 rate, 5.210961 seconds, 1664 images
Loaded: 0.000046 seconds

 53: 1814.482910, 1386.853760 avg loss, 0.000000 rate, 5.212034 seconds, 1696 images
Loaded: 0.000038 seconds

 54: 1809.125122, 1429.080933 avg loss, 0.000000 rate, 5.264531 seconds, 1728 images
Loaded: 0.000035 seconds

 55: 1799.300415, 1466.102905 avg loss, 0.000000 rate, 5.285175 seconds, 1760 images
Loaded: 0.000038 seconds

 56: 1790.291260, 1498.521729 avg loss, 0.000000 rate, 5.317025 seconds, 1792 images
Loaded: 0.000041 seconds

 57: 1783.576050, 1527.027100 avg loss, 0.000000 rate, 5.298457 seconds, 1824 images
Loaded: 0.000038 seconds

 58: 1769.597900, 1551.284180 avg loss, 0.000000 rate, 5.296273 seconds, 1856 images
Loaded: 0.000039 seconds

 59: 1758.136841, 1571.969482 avg loss, 0.000000 rate, 5.277847 seconds, 1888 images
Loaded: 0.000060 seconds

 60: 1745.222412, 1589.294800 avg loss, 0.000000 rate, 5.271202 seconds, 1920 images
Resizing
480 x 480 
 try to allocate additional workspace_size = 67.74 MB 
 CUDA allocate done! 
Loaded: 3.301125 seconds

 61: 1082.239624, 1538.589233 avg loss, 0.000000 rate, 3.138971 seconds, 1952 images
Loaded: 1.126361 seconds

 62: 1071.890503, 1491.919312 avg loss, 0.000000 rate, 3.092319 seconds, 1984 images
Loaded: 1.165808 seconds

 63: 1063.156494, 1449.043091 avg loss, 0.000000 rate, 3.109089 seconds, 2016 images
Loaded: 1.170386 seconds

 64: 1053.991577, 1409.537964 avg loss, 0.000000 rate, 3.103340 seconds, 2048 images
Loaded: 1.188800 seconds

 65: 1042.110718, 1372.795288 avg loss, 0.000000 rate, 3.103318 seconds, 2080 images
Loaded: 1.205747 seconds

 66: 1033.892700, 1338.905029 avg loss, 0.000000 rate, 3.134714 seconds, 2112 images
Loaded: 1.204720 seconds

 67: 1022.867065, 1307.301270 avg loss, 0.000000 rate, 3.124842 seconds, 2144 images
Loaded: 1.241954 seconds

 68: 1013.026917, 1277.873779 avg loss, 0.000000 rate, 3.124912 seconds, 2176 images
Loaded: 1.177054 seconds

 69: 999.903748, 1250.076782 avg loss, 0.000000 rate, 3.137247 seconds, 2208 images
Loaded: 1.195041 seconds

 70: 989.606567, 1224.029785 avg loss, 0.000000 rate, 3.144312 seconds, 2240 images
Resizing
480 x 480 
 try to allocate additional workspace_size = 67.74 MB 
 CUDA allocate done! 
Loaded: 3.594387 seconds

 71: 978.414246, 1199.468262 avg loss, 0.000000 rate, 3.201471 seconds, 2272 images
Loaded: 1.101660 seconds

 72: 964.543152, 1175.975708 avg loss, 0.000000 rate, 3.126491 seconds, 2304 images
Loaded: 1.192848 seconds

 73: 953.074524, 1153.685547 avg loss, 0.000000 rate, 3.121062 seconds, 2336 images
Loaded: 1.233922 seconds

 74: 940.636963, 1132.380737 avg loss, 0.000000 rate, 3.115576 seconds, 2368 images
Loaded: 1.182876 seconds

 75: 926.594055, 1111.802124 avg loss, 0.000000 rate, 3.130939 seconds, 2400 images
Loaded: 1.208158 seconds

 76: 915.166565, 1092.138550 avg loss, 0.000000 rate, 3.129372 seconds, 2432 images
Loaded: 1.206892 seconds

 77: 898.736938, 1072.798340 avg loss, 0.000000 rate, 3.166308 seconds, 2464 images
Loaded: 1.170312 seconds

 78: 885.443420, 1054.062866 avg loss, 0.000000 rate, 3.135046 seconds, 2496 images
Loaded: 1.210277 seconds

 79: 870.103638, 1035.666992 avg loss, 0.000000 rate, 3.135917 seconds, 2528 images
Loaded: 1.121796 seconds

 80: 854.039917, 1017.504272 avg loss, 0.000000 rate, 3.120471 seconds, 2560 images
Resizing
416 x 416 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 2.655243 seconds

 81: 631.755859, 978.929443 avg loss, 0.000000 rate, 2.612566 seconds, 2592 images
Loaded: 1.529265 seconds

 82: 616.779846, 942.714478 avg loss, 0.000000 rate, 2.524497 seconds, 2624 images
Loaded: 1.641470 seconds

 83: 607.249939, 909.168030 avg loss, 0.000000 rate, 2.542649 seconds, 2656 images
Loaded: 1.656410 seconds

 84: 595.012390, 877.752441 avg loss, 0.000000 rate, 2.530965 seconds, 2688 images
Loaded: 1.589436 seconds

 85: 581.292725, 848.106445 avg loss, 0.000000 rate, 2.529419 seconds, 2720 images
Loaded: 1.588214 seconds

 86: 571.060181, 820.401794 avg loss, 0.000000 rate, 2.559183 seconds, 2752 images
Loaded: 1.601548 seconds

 87: 559.255127, 794.287109 avg loss, 0.000000 rate, 2.567561 seconds, 2784 images
Loaded: 1.595382 seconds

 88: 546.810303, 769.539429 avg loss, 0.000000 rate, 2.564659 seconds, 2816 images
Loaded: 1.595492 seconds

 89: 532.396729, 745.825134 avg loss, 0.000000 rate, 2.564729 seconds, 2848 images
Loaded: 1.634707 seconds

 90: 521.072388, 723.349854 avg loss, 0.000000 rate, 2.561303 seconds, 2880 images
Resizing
384 x 384 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.650885 seconds

 91: 434.252502, 694.440125 avg loss, 0.000000 rate, 2.210872 seconds, 2912 images
Loaded: 1.906415 seconds

 92: 424.362640, 667.432373 avg loss, 0.000000 rate, 2.182973 seconds, 2944 images
Loaded: 1.936787 seconds

 93: 413.032623, 641.992371 avg loss, 0.000000 rate, 2.203982 seconds, 2976 images
Loaded: 1.885590 seconds

 94: 401.493866, 617.942505 avg loss, 0.000000 rate, 2.153415 seconds, 3008 images
Loaded: 1.892202 seconds

 95: 391.293915, 595.277649 avg loss, 0.000000 rate, 2.146084 seconds, 3040 images
Loaded: 1.884520 seconds

 96: 382.878326, 574.037720 avg loss, 0.000000 rate, 2.171469 seconds, 3072 images
Loaded: 1.870291 seconds

 97: 370.364380, 553.670410 avg loss, 0.000000 rate, 2.145424 seconds, 3104 images
Loaded: 1.990863 seconds

 98: 362.597534, 534.563110 avg loss, 0.000000 rate, 2.206789 seconds, 3136 images
Loaded: 1.881059 seconds

 99: 351.359680, 516.242798 avg loss, 0.000000 rate, 2.175958 seconds, 3168 images
Loaded: 1.891335 seconds

 100: 342.839050, 498.902435 avg loss, 0.000000 rate, 2.150638 seconds, 3200 images
Saving weights to backup/custom-train-yolo_last.weights
Resizing
608 x 608 
 try to allocate additional workspace_size = 108.69 MB 
 CUDA allocate done! 
Loaded: 1.456913 seconds

 101: 833.565125, 532.368713 avg loss, 0.000000 rate, 5.044236 seconds, 3232 images
Loaded: 0.000044 seconds

 102: 812.392700, 560.371094 avg loss, 0.000000 rate, 5.271101 seconds, 3264 images
Loaded: 0.000041 seconds

 103: 779.596924, 582.293701 avg loss, 0.000000 rate, 5.312495 seconds, 3296 images
Loaded: 0.000046 seconds

 104: 743.747925, 598.439148 avg loss, 0.000000 rate, 5.368146 seconds, 3328 images
Loaded: 0.000040 seconds

 105: 710.898804, 609.685120 avg loss, 0.000000 rate, 5.382980 seconds, 3360 images
Loaded: 0.000049 seconds

 106: 673.879822, 616.104614 avg loss, 0.000000 rate, 5.359816 seconds, 3392 images
Loaded: 0.000031 seconds

 107: 643.955200, 618.889648 avg loss, 0.000000 rate, 5.284211 seconds, 3424 images
Loaded: 0.000056 seconds

 108: 614.828735, 618.483582 avg loss, 0.000000 rate, 5.281869 seconds, 3456 images
Loaded: 0.000046 seconds

 109: 572.222168, 613.857422 avg loss, 0.000000 rate, 5.241708 seconds, 3488 images
Loaded: 0.000039 seconds

 110: 538.186462, 606.290344 avg loss, 0.000000 rate, 5.239917 seconds, 3520 images
Resizing
384 x 384 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 2.579256 seconds

 111: 200.320068, 565.693298 avg loss, 0.000000 rate, 2.193915 seconds, 3552 images
Loaded: 1.867938 seconds

 112: 187.832062, 527.907166 avg loss, 0.000000 rate, 2.145433 seconds, 3584 images
Loaded: 1.933409 seconds

 113: 177.113708, 492.827820 avg loss, 0.000000 rate, 2.235312 seconds, 3616 images
Loaded: 1.885198 seconds

 114: 167.025650, 460.247589 avg loss, 0.000000 rate, 2.203499 seconds, 3648 images
Loaded: 1.838439 seconds

 115: 155.230103, 429.745850 avg loss, 0.000000 rate, 2.210538 seconds, 3680 images
Loaded: 1.835176 seconds

 116: 148.263229, 401.597595 avg loss, 0.000000 rate, 2.203216 seconds, 3712 images
Loaded: 1.849283 seconds

 117: 140.183731, 375.456207 avg loss, 0.000000 rate, 2.150578 seconds, 3744 images
Loaded: 1.905455 seconds

 118: 132.190613, 351.129639 avg loss, 0.000000 rate, 2.258524 seconds, 3776 images
Loaded: 1.802250 seconds

 119: 124.438171, 328.460480 avg loss, 0.000000 rate, 2.195672 seconds, 3808 images
Loaded: 1.977947 seconds

 120: 118.055153, 307.419952 avg loss, 0.000000 rate, 2.270905 seconds, 3840 images
Resizing
384 x 384 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.523360 seconds

 121: 113.103004, 287.988251 avg loss, 0.000000 rate, 2.245847 seconds, 3872 images
Loaded: 1.865154 seconds

 122: 106.645790, 269.854004 avg loss, 0.000000 rate, 2.162207 seconds, 3904 images
Loaded: 1.924662 seconds

 123: 100.780869, 252.946686 avg loss, 0.000000 rate, 2.220473 seconds, 3936 images
Loaded: 1.910227 seconds

 124: 96.430603, 237.295074 avg loss, 0.000000 rate, 2.161338 seconds, 3968 images
Loaded: 1.913990 seconds

 125: 92.266151, 222.792175 avg loss, 0.000000 rate, 2.207975 seconds, 4000 images
Loaded: 1.822497 seconds

 126: 86.657715, 209.178726 avg loss, 0.000000 rate, 2.181968 seconds, 4032 images
Loaded: 1.850749 seconds

 127: 83.883553, 196.649216 avg loss, 0.000000 rate, 2.193679 seconds, 4064 images
Loaded: 1.834605 seconds

 128: 78.451416, 184.829437 avg loss, 0.000000 rate, 2.191081 seconds, 4096 images
Loaded: 1.908137 seconds

 129: 75.005966, 173.847092 avg loss, 0.000000 rate, 2.224768 seconds, 4128 images
Loaded: 1.873769 seconds

 130: 72.166557, 163.679031 avg loss, 0.000000 rate, 2.172353 seconds, 4160 images
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 1.792548 seconds

 131: 134.969208, 160.808044 avg loss, 0.000000 rate, 4.273266 seconds, 4192 images
Loaded: 0.236768 seconds

 132: 128.946609, 157.621902 avg loss, 0.000000 rate, 4.263710 seconds, 4224 images
Loaded: 0.200166 seconds

 133: 122.032188, 154.062927 avg loss, 0.000000 rate, 4.302868 seconds, 4256 images
Loaded: 0.160066 seconds

 134: 116.092636, 150.265900 avg loss, 0.000000 rate, 4.316426 seconds, 4288 images
Loaded: 0.242851 seconds

 135: 108.910820, 146.130386 avg loss, 0.000000 rate, 4.350989 seconds, 4320 images
Loaded: 0.388196 seconds

 136: 104.916748, 142.009018 avg loss, 0.000000 rate, 4.369681 seconds, 4352 images
Loaded: 0.244277 seconds

 137: 97.996490, 137.607758 avg loss, 0.000000 rate, 4.374640 seconds, 4384 images
Loaded: 0.313281 seconds

 138: 90.107605, 132.857742 avg loss, 0.000000 rate, 4.351158 seconds, 4416 images
Loaded: 0.146478 seconds

 139: 85.265114, 128.098480 avg loss, 0.000000 rate, 4.318348 seconds, 4448 images
Loaded: 0.355750 seconds

 140: 79.574203, 123.246056 avg loss, 0.000000 rate, 4.298348 seconds, 4480 images
Resizing
352 x 352 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.028057 seconds

 141: 32.627121, 114.184158 avg loss, 0.000000 rate, 1.940968 seconds, 4512 images
Loaded: 2.255652 seconds

 142: 30.734022, 105.839142 avg loss, 0.000000 rate, 1.959441 seconds, 4544 images
Loaded: 2.176846 seconds

 143: 30.411823, 98.296410 avg loss, 0.000000 rate, 1.992478 seconds, 4576 images
Loaded: 2.196345 seconds

 144: 28.920910, 91.358856 avg loss, 0.000000 rate, 1.952484 seconds, 4608 images
Loaded: 2.204387 seconds

 145: 28.131483, 85.036118 avg loss, 0.000000 rate, 1.953267 seconds, 4640 images
Loaded: 2.120347 seconds

 146: 25.420998, 79.074608 avg loss, 0.000000 rate, 1.943617 seconds, 4672 images
Loaded: 2.109047 seconds

 147: 24.353718, 73.602516 avg loss, 0.000000 rate, 1.963106 seconds, 4704 images
Loaded: 2.134952 seconds

 148: 22.634462, 68.505714 avg loss, 0.000000 rate, 1.992945 seconds, 4736 images
Loaded: 2.050731 seconds

 149: 21.825613, 63.837704 avg loss, 0.000000 rate, 2.028410 seconds, 4768 images
Loaded: 2.023473 seconds

 150: 21.272646, 59.581200 avg loss, 0.000001 rate, 1.973313 seconds, 4800 images
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 1.879389 seconds

 151: 43.145645, 57.937645 avg loss, 0.000001 rate, 4.394621 seconds, 4832 images
Loaded: 0.183937 seconds

 152: 40.459229, 56.189804 avg loss, 0.000001 rate, 4.341198 seconds, 4864 images
Loaded: 0.246996 seconds

 153: 38.034592, 54.374283 avg loss, 0.000001 rate, 4.387753 seconds, 4896 images
Loaded: 0.268229 seconds

 154: 36.227634, 52.559616 avg loss, 0.000001 rate, 4.385446 seconds, 4928 images
Loaded: 0.247426 seconds

 155: 34.744434, 50.778099 avg loss, 0.000001 rate, 4.400717 seconds, 4960 images
Loaded: 0.285536 seconds

 156: 33.606224, 49.060913 avg loss, 0.000001 rate, 4.398110 seconds, 4992 images
Loaded: 0.329813 seconds

 157: 31.150736, 47.269894 avg loss, 0.000001 rate, 4.337512 seconds, 5024 images
Loaded: 0.234568 seconds

 158: 29.132837, 45.456188 avg loss, 0.000001 rate, 4.331877 seconds, 5056 images
Loaded: 0.132989 seconds

 159: 28.311682, 43.741737 avg loss, 0.000001 rate, 4.287025 seconds, 5088 images
Loaded: 0.325663 seconds

 160: 26.235325, 41.991096 avg loss, 0.000001 rate, 4.264476 seconds, 5120 images
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 3.672322 seconds

 161: 24.264271, 40.218414 avg loss, 0.000001 rate, 4.264329 seconds, 5152 images
Loaded: 0.164521 seconds

 162: 23.592842, 38.555859 avg loss, 0.000001 rate, 4.236755 seconds, 5184 images
Loaded: 0.250285 seconds

 163: 21.418007, 36.842072 avg loss, 0.000001 rate, 4.248697 seconds, 5216 images
Loaded: 0.308233 seconds

 164: 19.679268, 35.125790 avg loss, 0.000001 rate, 4.264140 seconds, 5248 images
Loaded: 0.229786 seconds

 165: 18.242283, 33.437439 avg loss, 0.000001 rate, 4.278987 seconds, 5280 images
Loaded: 0.289195 seconds

 166: 18.094856, 31.903181 avg loss, 0.000001 rate, 4.316477 seconds, 5312 images
Loaded: 0.198735 seconds

 167: 16.608103, 30.373672 avg loss, 0.000001 rate, 4.330190 seconds, 5344 images
Loaded: 0.138024 seconds

 168: 15.586241, 28.894930 avg loss, 0.000001 rate, 4.379158 seconds, 5376 images
Loaded: 0.156272 seconds

 169: 14.441493, 27.449587 avg loss, 0.000001 rate, 4.341835 seconds, 5408 images
Loaded: 0.174819 seconds

 170: 14.212545, 26.125883 avg loss, 0.000001 rate, 4.328561 seconds, 5440 images
Resizing
352 x 352 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 2.935472 seconds

 171: 7.745799, 24.287874 avg loss, 0.000001 rate, 1.993777 seconds, 5472 images
Loaded: 2.086190 seconds

 172: 8.435947, 22.702682 avg loss, 0.000001 rate, 1.970425 seconds, 5504 images
Loaded: 2.110358 seconds

 173: 8.148928, 21.247307 avg loss, 0.000001 rate, 1.957353 seconds, 5536 images
Loaded: 2.058888 seconds

 174: 7.831917, 19.905767 avg loss, 0.000001 rate, 1.954481 seconds, 5568 images
Loaded: 2.046013 seconds

 175: 6.531989, 18.568390 avg loss, 0.000001 rate, 1.969262 seconds, 5600 images
Loaded: 2.129559 seconds

 176: 6.906528, 17.402205 avg loss, 0.000001 rate, 2.001699 seconds, 5632 images
Loaded: 1.983966 seconds

 177: 6.731628, 16.335146 avg loss, 0.000001 rate, 1.966610 seconds, 5664 images
Loaded: 2.051160 seconds

 178: 6.180530, 15.319684 avg loss, 0.000001 rate, 1.958165 seconds, 5696 images
Loaded: 2.103079 seconds

 179: 6.048562, 14.392571 avg loss, 0.000001 rate, 2.008537 seconds, 5728 images
Loaded: 2.083856 seconds

 180: 5.860236, 13.539338 avg loss, 0.000001 rate, 1.998765 seconds, 5760 images
Resizing
512 x 512 
 try to allocate additional workspace_size = 77.07 MB 
 CUDA allocate done! 
Loaded: 2.024241 seconds

 181: 7.802999, 12.965704 avg loss, 0.000001 rate, 3.483028 seconds, 5792 images
Loaded: 1.022089 seconds

 182: 7.911579, 12.460292 avg loss, 0.000001 rate, 3.412240 seconds, 5824 images
Loaded: 1.105981 seconds

 183: 7.377176, 11.951981 avg loss, 0.000001 rate, 3.413583 seconds, 5856 images
Loaded: 1.038228 seconds

 184: 7.362961, 11.493078 avg loss, 0.000001 rate, 3.437693 seconds, 5888 images
Loaded: 1.091333 seconds

 185: 6.977678, 11.041538 avg loss, 0.000001 rate, 3.427781 seconds, 5920 images
Loaded: 1.023922 seconds

 186: 7.140187, 10.651403 avg loss, 0.000001 rate, 3.418737 seconds, 5952 images
Loaded: 1.075415 seconds

 187: 6.691795, 10.255443 avg loss, 0.000001 rate, 3.408569 seconds, 5984 images
Loaded: 1.089609 seconds

 188: 6.532420, 9.883141 avg loss, 0.000001 rate, 3.414640 seconds, 6016 images
Loaded: 1.051980 seconds

 189: 6.869525, 9.581779 avg loss, 0.000001 rate, 3.405018 seconds, 6048 images
Loaded: 1.199142 seconds

 190: 6.749388, 9.298540 avg loss, 0.000001 rate, 3.375930 seconds, 6080 images
Resizing
512 x 512 
 try to allocate additional workspace_size = 77.07 MB 
 CUDA allocate done! 
Loaded: 3.658429 seconds

 191: 5.732538, 8.941940 avg loss, 0.000001 rate, 3.421550 seconds, 6112 images
Loaded: 1.115487 seconds

 192: 5.763366, 8.624083 avg loss, 0.000001 rate, 3.362478 seconds, 6144 images
Loaded: 1.232382 seconds

 193: 5.246841, 8.286359 avg loss, 0.000001 rate, 3.339203 seconds, 6176 images
Loaded: 1.142517 seconds

 194: 5.499609, 8.007684 avg loss, 0.000001 rate, 3.368222 seconds, 6208 images
Loaded: 1.133876 seconds

 195: 4.988072, 7.705723 avg loss, 0.000001 rate, 3.387010 seconds, 6240 images
Loaded: 1.150572 seconds

 196: 5.495774, 7.484728 avg loss, 0.000001 rate, 3.402709 seconds, 6272 images
Loaded: 1.113108 seconds

 197: 4.907601, 7.227015 avg loss, 0.000002 rate, 3.443113 seconds, 6304 images
Loaded: 1.120824 seconds

 198: 4.873627, 6.991676 avg loss, 0.000002 rate, 3.438920 seconds, 6336 images
Loaded: 1.136368 seconds

 199: 5.526850, 6.845194 avg loss, 0.000002 rate, 3.422021 seconds, 6368 images
Loaded: 1.133953 seconds

 200: 4.459447, 6.606619 avg loss, 0.000002 rate, 3.424253 seconds, 6400 images
Saving weights to backup/custom-train-yolo_last.weights
Resizing
544 x 544 
 try to allocate additional workspace_size = 87.01 MB 
 CUDA allocate done! 
Loaded: 1.277495 seconds

 201: 4.939904, 6.439948 avg loss, 0.000002 rate, 4.332464 seconds, 6432 images
Loaded: 0.381301 seconds

 202: 4.862922, 6.282245 avg loss, 0.000002 rate, 4.275336 seconds, 6464 images
Loaded: 0.438626 seconds

 203: 4.085641, 6.062585 avg loss, 0.000002 rate, 4.296002 seconds, 6496 images
Loaded: 0.434540 seconds

 204: 4.627556, 5.919082 avg loss, 0.000002 rate, 4.313246 seconds, 6528 images
Loaded: 0.468392 seconds

 205: 4.200765, 5.747251 avg loss, 0.000002 rate, 4.321316 seconds, 6560 images
Loaded: 0.424981 seconds

 206: 4.451098, 5.617635 avg loss, 0.000002 rate, 4.323214 seconds, 6592 images
Loaded: 0.446376 seconds

 207: 4.543732, 5.510245 avg loss, 0.000002 rate, 4.340714 seconds, 6624 images
Loaded: 0.372179 seconds

 208: 3.909205, 5.350141 avg loss, 0.000002 rate, 4.316957 seconds, 6656 images
Loaded: 0.403008 seconds

 209: 3.634208, 5.178548 avg loss, 0.000002 rate, 4.311997 seconds, 6688 images
Loaded: 0.394218 seconds

 210: 3.813939, 5.042087 avg loss, 0.000002 rate, 4.294049 seconds, 6720 images
Resizing
480 x 480 
 try to allocate additional workspace_size = 67.74 MB 
 CUDA allocate done! 
Loaded: 3.749630 seconds

 211: 3.707102, 4.908588 avg loss, 0.000002 rate, 3.155570 seconds, 6752 images
Loaded: 1.360379 seconds

 212: 3.545493, 4.772279 avg loss, 0.000002 rate, 3.085527 seconds, 6784 images
Loaded: 1.362915 seconds

 213: 3.313728, 4.626424 avg loss, 0.000002 rate, 3.098091 seconds, 6816 images
Loaded: 1.372935 seconds

 214: 3.504254, 4.514207 avg loss, 0.000002 rate, 3.131396 seconds, 6848 images
Loaded: 1.366477 seconds

 215: 3.435507, 4.406337 avg loss, 0.000002 rate, 3.156800 seconds, 6880 images
Loaded: 1.334902 seconds

 216: 3.275230, 4.293226 avg loss, 0.000002 rate, 3.140537 seconds, 6912 images
Loaded: 1.384648 seconds

 217: 3.738048, 4.237709 avg loss, 0.000002 rate, 3.137388 seconds, 6944 images
Loaded: 1.367137 seconds

 218: 3.232055, 4.137143 avg loss, 0.000002 rate, 3.154233 seconds, 6976 images
Loaded: 1.359376 seconds

 219: 3.323618, 4.055791 avg loss, 0.000002 rate, 3.132223 seconds, 7008 images
Loaded: 1.496647 seconds

 220: 3.472374, 3.997449 avg loss, 0.000002 rate, 3.134573 seconds, 7040 images
Resizing
416 x 416 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.706855 seconds

 221: 3.714732, 3.969177 avg loss, 0.000002 rate, 2.623176 seconds, 7072 images
Loaded: 1.757894 seconds

 222: 4.011533, 3.973413 avg loss, 0.000002 rate, 2.525336 seconds, 7104 images
Loaded: 1.934961 seconds

 223: 3.736531, 3.949725 avg loss, 0.000002 rate, 2.620534 seconds, 7136 images
Loaded: 1.762234 seconds

 224: 3.347547, 3.889507 avg loss, 0.000003 rate, 2.551301 seconds, 7168 images
Loaded: 1.721676 seconds

 225: 3.187762, 3.819332 avg loss, 0.000003 rate, 2.549750 seconds, 7200 images
Loaded: 1.781864 seconds

 226: 2.545011, 3.691900 avg loss, 0.000003 rate, 2.567164 seconds, 7232 images
Loaded: 1.770716 seconds

 227: 2.828764, 3.605587 avg loss, 0.000003 rate, 2.556094 seconds, 7264 images
Loaded: 1.706427 seconds

 228: 3.269938, 3.572022 avg loss, 0.000003 rate, 2.571266 seconds, 7296 images
Loaded: 1.685629 seconds

 229: 3.260827, 3.540902 avg loss, 0.000003 rate, 2.549249 seconds, 7328 images
Loaded: 1.729444 seconds

 230: 3.503815, 3.537194 avg loss, 0.000003 rate, 2.543537 seconds, 7360 images
Resizing
512 x 512 
 try to allocate additional workspace_size = 77.07 MB 
 CUDA allocate done! 
Loaded: 2.639223 seconds

 231: 2.999933, 3.483467 avg loss, 0.000003 rate, 3.404676 seconds, 7392 images
Loaded: 1.030125 seconds

 232: 3.210326, 3.456153 avg loss, 0.000003 rate, 3.371518 seconds, 7424 images
Loaded: 1.101046 seconds

 233: 2.997488, 3.410287 avg loss, 0.000003 rate, 3.383016 seconds, 7456 images
Loaded: 1.108105 seconds

 234: 2.740766, 3.343335 avg loss, 0.000003 rate, 3.375398 seconds, 7488 images
Loaded: 1.110328 seconds

 235: 2.772891, 3.286290 avg loss, 0.000003 rate, 3.425432 seconds, 7520 images
Loaded: 1.084188 seconds

 236: 3.201218, 3.277783 avg loss, 0.000003 rate, 3.457361 seconds, 7552 images
Loaded: 1.046970 seconds

 237: 2.827726, 3.232777 avg loss, 0.000003 rate, 3.408544 seconds, 7584 images
Loaded: 1.078200 seconds

 238: 2.811347, 3.190634 avg loss, 0.000003 rate, 3.417175 seconds, 7616 images
Loaded: 1.133859 seconds

 239: 2.886034, 3.160174 avg loss, 0.000003 rate, 3.398966 seconds, 7648 images
Loaded: 1.157345 seconds

 240: 2.469579, 3.091115 avg loss, 0.000003 rate, 3.380106 seconds, 7680 images
Resizing
480 x 480 
 try to allocate additional workspace_size = 67.74 MB 
 CUDA allocate done! 
Loaded: 3.678301 seconds

 241: 2.766567, 3.058660 avg loss, 0.000003 rate, 3.154204 seconds, 7712 images
Loaded: 1.348319 seconds

 242: 2.560811, 3.008875 avg loss, 0.000003 rate, 3.108731 seconds, 7744 images
Loaded: 1.408724 seconds

 243: 2.804442, 2.988432 avg loss, 0.000003 rate, 3.090909 seconds, 7776 images
Loaded: 1.444137 seconds

 244: 2.937446, 2.983333 avg loss, 0.000004 rate, 3.113725 seconds, 7808 images
Loaded: 1.313836 seconds

 245: 2.460112, 2.931011 avg loss, 0.000004 rate, 3.105529 seconds, 7840 images
Loaded: 1.361812 seconds

 246: 2.987369, 2.936647 avg loss, 0.000004 rate, 3.122869 seconds, 7872 images
Loaded: 1.341628 seconds

 247: 2.755343, 2.918516 avg loss, 0.000004 rate, 3.121588 seconds, 7904 images
Loaded: 1.252970 seconds

 248: 2.769065, 2.903571 avg loss, 0.000004 rate, 3.120810 seconds, 7936 images
Loaded: 1.342613 seconds

 249: 2.680665, 2.881281 avg loss, 0.000004 rate, 3.139040 seconds, 7968 images
Loaded: 1.315283 seconds

 250: 2.987723, 2.891925 avg loss, 0.000004 rate, 3.131990 seconds, 8000 images
Resizing
320 x 320 
 try to allocate additional workspace_size = 52.43 MB 
 CUDA allocate done! 
Loaded: 3.095017 seconds

 251: 2.994536, 2.902186 avg loss, 0.000004 rate, 1.775438 seconds, 8032 images
Loaded: 2.364468 seconds

 252: 2.839226, 2.895890 avg loss, 0.000004 rate, 1.749736 seconds, 8064 images
Loaded: 2.363665 seconds

 253: 3.125005, 2.918802 avg loss, 0.000004 rate, 1.776987 seconds, 8096 images
Loaded: 2.448214 seconds

 254: 3.290020, 2.955923 avg loss, 0.000004 rate, 1.780010 seconds, 8128 images
Loaded: 2.409418 seconds

 255: 2.987749, 2.959106 avg loss, 0.000004 rate, 1.783683 seconds, 8160 images
Loaded: 2.428998 seconds

 256: 3.023077, 2.965503 avg loss, 0.000004 rate, 1.797217 seconds, 8192 images
Loaded: 2.387981 seconds

 257: 2.410569, 2.910010 avg loss, 0.000004 rate, 1.771482 seconds, 8224 images
Loaded: 2.359795 seconds

 258: 3.296492, 2.948658 avg loss, 0.000004 rate, 1.783732 seconds, 8256 images
Loaded: 2.348948 seconds

 259: 3.565882, 3.010381 avg loss, 0.000004 rate, 1.768584 seconds, 8288 images
Loaded: 2.386864 seconds

 260: 3.511684, 3.060511 avg loss, 0.000005 rate, 1.765717 seconds, 8320 images
Resizing
512 x 512 
 try to allocate additional workspace_size = 77.07 MB 
 CUDA allocate done! 
Loaded: 2.173649 seconds

 261: 2.464088, 3.000869 avg loss, 0.000005 rate, 3.447791 seconds, 8352 images
Loaded: 1.184983 seconds

 262: 2.144822, 2.915264 avg loss, 0.000005 rate, 3.406516 seconds, 8384 images
Loaded: 1.219133 seconds

 263: 2.565091, 2.880247 avg loss, 0.000005 rate, 3.440608 seconds, 8416 images
Loaded: 1.178984 seconds

 264: 2.503429, 2.842565 avg loss, 0.000005 rate, 3.419891 seconds, 8448 images
Loaded: 1.188001 seconds

 265: 2.473848, 2.805693 avg loss, 0.000005 rate, 3.445765 seconds, 8480 images
Loaded: 1.196070 seconds

 266: 1.975374, 2.722661 avg loss, 0.000005 rate, 3.429824 seconds, 8512 images
Loaded: 1.293797 seconds

 267: 2.557600, 2.706155 avg loss, 0.000005 rate, 3.407241 seconds, 8544 images
Loaded: 1.227031 seconds
In [29]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_last.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_last.weights...Couldn't open file: backup/custom-train-yolo_last.weights
In [30]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_final.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_final.weights...Couldn't open file: backup/custom-train-yolo_final.weights
In [31]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_4000.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_4000.weights...Couldn't open file: backup/custom-train-yolo_4000.weights
In [32]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_3000.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_3000.weights...Couldn't open file: backup/custom-train-yolo_3000.weights
In [22]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_2000.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_2000.weights...Couldn't open file: backup/custom-train-yolo_2000.weights
In [25]:
!./darknet detector map custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_1000.weights

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_1000.weights...Couldn't open file: backup/custom-train-yolo_1000.weights
In [24]:
if os.path.exists('./backup'):
    !cp -r ./backup "/content/drive/MyDrive/darknet"  

if not os.path.exists('./backup'):
    os.makedirs('./backup')
    !cp -r "/content/drive/MyDrive/darknet/backup/custom-train-yolo_final.weights" ./backup

cp: cannot stat '/content/drive/MyDrive/darknet/backup/custom-train-yolo_final.weights': No such file or directory
In [26]:
!ls -al ./backup

total 8
drwxr-xr-x 2 root root 4096 Dec 22 23:20 .
drwxr-xr-x 7 root root 4096 Dec 22 23:20 ..
In [27]:
!./darknet detector test custom/custom_data.data custom/custom-train-yolo.cfg backup/custom-train-yolo_final.weights data/fruit10.jpg -dont-show

imShow('predictions.jpg')

layer     filters    size              input                output
   0 conv     32  3 x 3 / 1   416 x 416 x   3   ->   416 x 416 x  32 0.299 BF
   1 conv     64  3 x 3 / 2   416 x 416 x  32   ->   208 x 208 x  64 1.595 BF
   2 conv     32  1 x 1 / 1   208 x 208 x  64   ->   208 x 208 x  32 0.177 BF
   3 conv     64  3 x 3 / 1   208 x 208 x  32   ->   208 x 208 x  64 1.595 BF
   4 Shortcut Layer: 1
   5 conv    128  3 x 3 / 2   208 x 208 x  64   ->   104 x 104 x 128 1.595 BF
   6 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
   7 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
   8 Shortcut Layer: 5
   9 conv     64  1 x 1 / 1   104 x 104 x 128   ->   104 x 104 x  64 0.177 BF
  10 conv    128  3 x 3 / 1   104 x 104 x  64   ->   104 x 104 x 128 1.595 BF
  11 Shortcut Layer: 8
  12 conv    256  3 x 3 / 2   104 x 104 x 128   ->    52 x  52 x 256 1.595 BF
  13 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  14 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  15 Shortcut Layer: 12
  16 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  17 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  18 Shortcut Layer: 15
  19 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  20 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  21 Shortcut Layer: 18
  22 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  23 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  24 Shortcut Layer: 21
  25 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  26 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  27 Shortcut Layer: 24
  28 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  29 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  30 Shortcut Layer: 27
  31 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  32 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  33 Shortcut Layer: 30
  34 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
  35 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
  36 Shortcut Layer: 33
  37 conv    512  3 x 3 / 2    52 x  52 x 256   ->    26 x  26 x 512 1.595 BF
  38 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  39 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  40 Shortcut Layer: 37
  41 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  42 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  43 Shortcut Layer: 40
  44 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  45 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  46 Shortcut Layer: 43
  47 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  48 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  49 Shortcut Layer: 46
  50 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  51 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  52 Shortcut Layer: 49
  53 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  54 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  55 Shortcut Layer: 52
  56 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  57 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  58 Shortcut Layer: 55
  59 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  60 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  61 Shortcut Layer: 58
  62 conv   1024  3 x 3 / 2    26 x  26 x 512   ->    13 x  13 x1024 1.595 BF
  63 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  64 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  65 Shortcut Layer: 62
  66 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  67 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  68 Shortcut Layer: 65
  69 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  70 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  71 Shortcut Layer: 68
  72 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  73 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  74 Shortcut Layer: 71
  75 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  76 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  77 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  78 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  79 conv    512  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x 512 0.177 BF
  80 conv   1024  3 x 3 / 1    13 x  13 x 512   ->    13 x  13 x1024 1.595 BF
  81 conv     21  1 x 1 / 1    13 x  13 x1024   ->    13 x  13 x  21 0.007 BF
  82 yolo
  83 route  79
  84 conv    256  1 x 1 / 1    13 x  13 x 512   ->    13 x  13 x 256 0.044 BF
  85 upsample            2x    13 x  13 x 256   ->    26 x  26 x 256
  86 route  85 61
  87 conv    256  1 x 1 / 1    26 x  26 x 768   ->    26 x  26 x 256 0.266 BF
  88 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  89 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  90 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  91 conv    256  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x 256 0.177 BF
  92 conv    512  3 x 3 / 1    26 x  26 x 256   ->    26 x  26 x 512 1.595 BF
  93 conv     21  1 x 1 / 1    26 x  26 x 512   ->    26 x  26 x  21 0.015 BF
  94 yolo
  95 route  91
  96 conv    128  1 x 1 / 1    26 x  26 x 256   ->    26 x  26 x 128 0.044 BF
  97 upsample            2x    26 x  26 x 128   ->    52 x  52 x 128
  98 route  97 36
  99 conv    128  1 x 1 / 1    52 x  52 x 384   ->    52 x  52 x 128 0.266 BF
 100 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 101 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 102 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 103 conv    128  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x 128 0.177 BF
 104 conv    256  3 x 3 / 1    52 x  52 x 128   ->    52 x  52 x 256 1.595 BF
 105 conv     21  1 x 1 / 1    52 x  52 x 256   ->    52 x  52 x  21 0.029 BF
 106 yolo
Total BFLOPS 65.297 
 Allocate additional workspace_size = 52.43 MB 
Loading weights from backup/custom-train-yolo_final.weights...Couldn't open file: backup/custom-train-yolo_final.weights
저작자표시 비영리 변경금지

'OpenCV > 머신비전 - 이미지 디텍팅' 카테고리의 다른 글

YOLO (gum, milk, chocolate) 추가 자료  (0) 2020.12.24
YOLO (gum, chocolate, milk / 실습)  (0) 2020.12.24
YOLO에 관한 이론  (0) 2020.12.24
labelImg  (0) 2020.12.24
YOLO.py  (0) 2020.12.23
'OpenCV/머신비전 - 이미지 디텍팅' Related Articles more
Comments