Home

Awesome

DRNet for Video Indvidual Counting (CVPR 2022)

Introduction

This is the official PyTorch implementation of paper: DR.VIC: Decomposition and Reasoning for Video Individual Counting. Different from the single image counting methods, it counts the total number of the pedestrians in a video sequence with a person in different frames only being calculated once. DRNet decomposes this new task to estimate the initial crowd number in the first frame and integrate differential crowd numbers in a set of following image pairs (namely current frame and preceding frame). framework

Catalog

Getting started

preparatoin

Training

Check some parameters in config.py before training,

Check other parameters (TRAIN_BATCH_SIZE, TRAIN_SIZE etc.) in the Root/DRNet/datasets/setting in case your GPU's memory is not support for the default setting.

Tips: The training process takes ~10 hours on HT21 dataset with one TITAN RTX (24GB Memory).

Testing

To reproduce the performance, download the pre-trained models from onedrive or badu disk and then place pretrained_models folder to Root/DRNet/model/

Performance

The results on HT21 and SenseCrowd.

MethodCroHD11~CroHD15MAE/MSE/MRAE(%)
Paper: VGG+FPN [2,3]164.6/1075.5/752.8/784.5/382.3141.1/192.3/27.4
This Repo's Reproduction: VGG+FPN [2,3]138.4/1017.5/623.9/659.8/348.5160.7/217.3/25.1
MethodMAE/MSE/MRAE(%)MIAE/MOAED0~D4 (for MAE)
Paper: VGG+FPN [2,3]12.3/24.7/12.71.98/2.014.1/8.0/23.3/50.0/77.0
This Repo's Reproduction: VGG+FPN [2,3]11.7/24.6/11.71.99/1.883.6/6.8/22.4/42.6/85.2

Video Demo

Please visit bilibili or YouTube to watch the video demonstration. demo

References

  1. Acquisition of Localization Confidence for Accurate Object Detection, ECCV, 2018.
  2. Very Deep Convolutional Networks for Large-scale Image Recognition, arXiv, 2014.
  3. Feature Pyramid Networks for Object Detection, CVPR, 2017.

Citation

If you find this project is useful for your research, please cite:

@article{han2022drvic,
  title={DR.VIC: Decomposition and Reasoning for Video Individual Counting},
  author={Han, Tao, Bai Lei, Gao, Junyu, Qi Wang, and Ouyang  Wanli},
  booktitle={CVPR},
  year={2022}
}

Acknowledgement

The released PyTorch training script borrows some codes from the C^3 Framework and SuperGlue repositories. If you think this repo is helpful for your research, please consider cite them.