Computer Vision

This is the final project of CSE204 Machine Learning course in Ecole Polytechnique. We use CNN and EfficientNet model to do OCR for singular handwriting Chinese Characters. The project is mainly based on Tensorflow 2.0 API.

The goal of this project is to develop a lossy image compression format close to JPEG. Lossy compression algorithms allow to greatly reduce the size of image files at the price of loosing some data from the original image.