Skip to content

Latest commit

 

History

History
109 lines (67 loc) · 2.83 KB

README.md

File metadata and controls

109 lines (67 loc) · 2.83 KB

[CVPRW2023] "Leveraging Future Trajectory Prediction for Multi-Camera People Tracking"

Track1: Multi-Camera People Tracking

The official repository for 7th NVIDIA AI City Challenge

Pipeline Overview

Environment

We run on 2 NVIDIA A6000 GPUs.

  • Linux or macOS
  • Python 3.7+ (Python 3.8 in our envs)
  • PyTorch 1.9+ (1.11.0 in our envs)
  • CUDA 10.2+ (CUDA 11.3 in our envs)
  • mmcv-full==1.7.1 (MMCV)

Installation

  • Step #1. Create environment (recommend environment)
conda env create --file environment.yaml
conda activate scit
  • Step #2. Install packages
sh setup.sh

Train

Object Detection
Keypoint Detection
  • Pretrained

    We directly use yolov7-pose-estimation's pretrained pose estimation model.

    You can download pretrained pose estimation model from their git page.

Trajectory Prediction

Demo

  • Step #1. Single-Camera Tracking.
sh run_scmt.sh
sh run_mcmt.sh

Citation

@InProceedings{Jeon_2023_CVPR,
    author    = {Jeon, Yuntae and Tran, Dai Quoc and Park, Minsoo and Park, Seunghee},
    title     = {Leveraging Future Trajectory Prediction for Multi-Camera People Tracking},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
    month     = {June},
    year      = {2023},
    pages     = {5398-5407}
}