DDAD: Detachable Crowd Density Estimation Assisted Pedestrian Detection

This is the pytorch implementation of our paper DDAD: Detachable Crowd Density Estimation Assisted Pedestrian Detection, published in IEEE Transactions on Intelligent Transportation Systems (T-ITS) 2023.

Quick start

Installation

• Install dependencies:

  pip install -r requirements.txt

• Training. More training seetings can be set in config.py

  cd tools
  python train.py -md rcnn_fpn_baseline
  python train.py -md rcnn_emd_simple_idad

• Testing.

  cd tools
  python test.py -md rcnn_fpn_baseline -r 1
  python test.py -md rcnn_emd_simple_idad -r 1

  When testing different models, simply modify the numbers in the test commands. The result json file will be evaluated automatically.

Models are avaliable in the model zoo, code: 4i2x.

Mat files are in the mat files, code: fpga.

BibTex

This is the repository for the paper MarginMatch: DDAD: Detachable Crowd Density Estimation Assisted Pedestrian Detection. If you found this repository helpful, consider citing our paper:

@article{tang2022ddad,
  title={DDAD: Detachable Crowd Density Estimation Assisted Pedestrian Detection},
  author={Tang, Wenxiao and Liu, Kun and Shakeel, M Saad and Wang, Hao and Kang, Wenxiong},
  journal={IEEE Transactions on Intelligent Transportation Systems},
  volume={24},
  number={2},
  pages={1867--1878},
  year={2022},
  publisher={IEEE}
}

Acknowledgement

• This code was inspired from CrowdDet.