This repository is based on the code for CenterTrack which can be found in the original repository: https://github.com/xingyizhou/CenterTrack
The DCNv2 code is adopted from https://github.com/MatthewHowe/DCNv2
This repository contains code and data for the 2D proximity and completeness fields from: https://github.com/Lijun-Yu/zero_virus
For setup clone the repository and install the conda environment:
git clone https://github.com/kocurvik/aicc
cd aicc
conda env create -f environment.ymlActivate the environment and then setup the DCNv2:
conda activate aicc
cd model/networks/DCNv2
python setup.py build developThe model should be downloaded from the CenterTrack model zoo. Specifically the MS COCO tracking model. Place the coco_tracking.pth file into the checkpoints directory.
To run on a single video use:
python run_single_video_threaded.py /path/to/video.mp4You can also try to use the serial version (slower):
python run_single_video_threaded.py /path/to/video.mp4To run the on the whole dataset in batched form run:
python run_batch_threaded.py -b 2 /path/to/AIC21_Track1_Vehicle_Counting/Dataset_AIf you find this code useful please consider citing our work:
@inproceedings{aicc2021comenius,
title={Multi-Class Multi-Movement Vehicle Counting Based on CenterTrack},
author={Kocur, Viktor and Ft\'{a}\v{c}nik, Milan},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
year={2021},
month={June},
pages = {4009-4015}
}