CSRNet-pytorch

This is a PyTorch Lightning implementation of CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes in CVPR 2018 which is one of the most popular crowd counting methods. The original implementation is here.

Datasets

ShanghaiTech Dataset: Google Drive

Prerequisites

We strongly recommend Anaconda as the environment.

Python: 3.8.13

PyTorch: 1.13.1

CUDA: 11.7

Ground Truth

An example of how to generate the density maps can be found in the density_map_generation.py file.

Training Process

Run python train.py to start training process.

Testing

python test.py --model [path to model] 

Results

To be added soon.

References

If you find this implementation useful, please give stars and cite the original paper:

@inproceedings{li2018csrnet,
  title={CSRNet: Dilated convolutional neural networks for understanding the highly congested scenes},
  author={Li, Yuhong and Zhang, Xiaofan and Chen, Deming},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={1091--1100},
  year={2018}
}

Please cite the Shanghai datasets and other works if you use them.

@inproceedings{zhang2016single,
  title={Single-image crowd counting via multi-column convolutional neural network},
  author={Zhang, Yingying and Zhou, Desen and Chen, Siqin and Gao, Shenghua and Ma, Yi},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={589--597},
  year={2016}
}