Shadow Removal via Shadow Decomposition

This project aims to use deep learning method for shadow removal. Inspired by physical models of shadow formation, they use a linear illumination transformation to model the shadow effects in the image that allows the shadow image to be expressed as a combination of the shadow-free image, the shadow parameters, and a matte layer. The model uses two deep networks, namely SP-Net and M-Net, to predict the shadow parameters and the shadow matte respectively.

• Drive Link - https://drive.google.com/drive/folders/19F7Q3jUY1k_FubXtcR1aimDrTSLtWpK7?usp=sharing

Reference work :

Le, H., & Samaras, D. (2019). Shadow removal via shadow image decomposition. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 8578-8587).

Instruction to project :

The python file train.py under the src/model directory contains the training code. We use VGG16 and UNet-256 for our training over the images. To run inference use the inference.py file also under the similar dir using the command :

python3 inference.py

The train.py file also holds the option() function which can be used to change the model parameters.

Results :

• Late days used 3.