/Boosting-Crowd-Counting-via-Multifaceted-Attention

Official Implement of CVPR 2022 paper 'Boosting Crowd Counting via Multifaceted Attention'

Primary LanguagePythonMIT LicenseMIT

Boosting-Crowd-Counting-via-Multifaceted-Attention

Official Implement of CVPR 2022 paper 'Boosting Crowd Counting via Multifaceted Attention'

arxiv | 知乎 | B站

image

Train

  1. Dowload Dataset JHU++ or UCF-QNRF.
  2. Preprocess them by 'preprocess_dataset.py' or 'preprocess_dataset_ucf.py'.
  3. Change the path to where your data and models are located in 'Train.py'.
  4. Run 'Train.py'
  5. Wait patiently and happily for the program to finish.
  6. Then you will get a good counting model!

Test

  1. Dowload Dataset JHU++ or UCF-QNRF.
  2. Preprocess them by 'preprocess_dataset.py' or 'preprocess_dataset_ucf.py'.
  3. JHU Model Link; UCF Model Link
  4. Change the path to where your data and models are located in 'Test.py'.
  5. Run 'Test.py'.

Citation

If you use this code for your research, please cite our paper:

@inproceedings{lin2022boosting,
  title={Boosting Crowd Counting via Multifaceted Attention},
  author={Lin, Hui and Ma, Zhiheng and Ji, Rongrong and Wang, Yaowei and Hong, Xiaopeng},
  booktitle={CVPR},
  year={2022}
}