The project is implementation of: https://arxiv.org/abs/1802.10062
The architecture is same as proposed in the paper but insted of ReLU i tried to make it LeakyReLU to prevent dead-ReLU problem
Official Implementation: https://github.com/leeyeehoo/CSRNet-pytorch
The Dataset Used: https://www.kaggle.com/datasets/tthien/shanghaitech-with-people-density-map
The model trained weights: https://drive.google.com/file/d/1uMcKoHd685MTJmdeK2uVmQvazkHiWtXE/view?usp=share_link