Train yolov3 on colab (I)

use Colab The benefits of ：
Computer building cuda The environment needs to have nvidia The graphics card （ I don't have ）, But in colab There is no need to build an environment , And it's free , Very friendly .

step ：

1、 Log in to Google hard disk ：https://drive.google.com
2、 Right click in the blank space of Google hard disk , Choose more -> Connect more applications ;

3、 Enter in the search box cloaboratory, Find the software and install .

4、 Get into Colab, Back to Google hard drive section , Click the right mouse button in the blank area again , Choose more , At this time, we found that Colab Is already installed , Open the can .

5、 Open the interface as follows , And click the connection in the upper right corner to wait for the connection to succeed .

6、 Upper left corner modification , Open notebook settings , choice GPU.

7、 Click in the following order , In the pop-up window, select connect to Google Hard disk in the cloud , The file directory appears on the left drive.

appear drive

8、 Click on “+” Add code , Download the sample program , Point triangle operation .

!git clone https://github.com/AlexeyAB/darknet

9、 Configure it

%cd darknet
!sed -i 's/OPENCV=0/OPENCV=1/' Makefile
!sed -i 's/GPU=0/GPU=1/' Makefile
!sed -i 's/CUDNN=0/CUDNN=1/' Makefile

10、 compile （ It's a little slow , It takes about a minute ）

!make

11、 download coco Data and pre training weight （ We need to make sure that the network is unblocked , Otherwise it will be very slow ）.

!wget https://pjreddie.com/media/files/yolov3.weights

12、 Add display 、 Upload 、 Download file code .

#  Show 
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.imshow(cv2.cvtColor(resized_image, cv2.COLOR_BGR2RGB))
  plt.show()

#  Upload files 
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)

#  Download the file  
def download(path):
  from google.colab import files
  files.download(path)

13、 All the work has been finished here . Check the environment, then use the official photos to test and show （ The training process is slow , Just wait patiently ）.

%cd /content/darknet
!./darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg
imShow('predictions.jpg')

give the result as follows ：

Up to now, it has been completed , If you want to train your pictures , You can go on and look down .

14、 Run the following code , Then upload your pictures to the disk where you have pictures

%cd ..
upload()

After uploading ：

15、 Run the load code , Change the picture name to your own .

%cd darknet
!./darknet detect cfg/yolov3.cfg yolov3.weights ../3.jpg
imShow('predictions.jpg')

16、 The effect shows ：
This is the original ：

This is the output diagram ：

complete .