We provide a PyTorch implementation our improved CSRNet which is based on the Bayesian crowd counting, called CSRNet.
For dataset and pretrained models refer to our paper.
python preprocess_dataset.py --origin_dir <directory of original data> --data_dir <directory of processed data>
python train.py --data_dir <directory of processed data> --save_dir <directory of log and model>
python test.py --data_dir <directory of processed data> --save_dir <directory of log and model>
This project is modified from Bayesian Crowd Counting project and is originally licensed under MIT | Copyright (c) 2020 Zhiheng_Ma.