The first step is to upload your images into your Google Drive in the folder YOLOv3, and then just run the Colab Notebook from the repository –—Panda.git

3. Train the Image dataset online

Train_Yolov3.ipynb is the training file – Setup like the above

Then run the model, you need not worry about training, and also each instruction to train the model is given in the Colab Notebook

Extracting the images
If you see an output similar to the one below, then well done.

Usually, it takes a minimum of 3 hours to the entire 12 hours you have the access to the GPU.

4. Test the model that we created

Every, 100 iterations, our custom object detector is going to be updated and saved on our Google Drive, inside the folder “yolov3”.

The file that we need is “yolov3_training_last.weights”.
You will find some other files that were saved on your drive, “yolov3_training__1000.weights”, “yolov3_training_2000.weights” and so on. The darknet makes a backup of the model for every 1000 iterations..

Test run 1

Yes, I was wondered – there was no inference😔

Test run 2 – After leaving the model to train for like 6 hours

Voila! – We got the Panda🐼😃

But, if you see the 5th tile, the result was scary, It is because of the disadvantage of Yolov3, and it’s limited to one class.

That’s it. The Object Detection is now over.

Real-time Applications of Object Detection

  1. Crowd counting
  2. Self-driving cars
  3. Video surveillance
  4. Face detection
  5. Anomaly detection
  6. You could try whatever you want

Everything in this link ->

This set of series will continue on multi-class object detection in the upcoming parts. 

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s