4g / unacademy

Background removal for Unacademy

We have 3 ways to remove the background
1. Browser side tensorflow.js
2. Server side opencv for plain background removal
3. Server side trained model for complex background removal

Run the tensorflow.js model

Try the browser side tensorflow.js live on http://bit.ly/unaback

The description and install of pretrianed tensorflow.js model

Checkout https://github.com/4g/unacademy/tree/master/tfjs-models

How to build and run the open cv

1] git clone this repository 2] Change to opencv code directory 3] Install opencv by pip opencv-python or other methods for different languages 4] Install python flask by pip install flask 5] flask run server.py 6] flask run client.py 7] Access the browser url of flask

Opencv description

1] It uses inrange function to pick particular color by hsv model 2] Dilate and erode 3] We find contours of lesser area comapared to the total image area. We then masked them.