MMPedestron Dataset Overview

Dataset Introduction

MMPedestron is a pedestrian detection dataset for multimodal learning, jointly developed by the following authors:

• Yi Zhang
• Wang ZENG
• Sheng Jin
• Chen Qian
• Ping Luo
• Wentao Liu

Dataset Configurations and Models

The dataset comprises multiple subsets and corresponding model configurations, as outlined below:

Region Proposal Performance

1. Pre‑training Stage

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

2. CrowdHuman

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

3. COCO‑Person

   • Method & Config: MMPedestron finetune
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

4. FLIR

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

5. PEDRo

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

6. LLVIP Datasets

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: Google Drive, Baidu Yun (Code: mmpd)

7. InOutDoor Datasets

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: same as above

8. STCrowd Datasets

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: same as above

9. EventPed Datasets

   • Method & Config: MMPedestron
   • Backbone: UNIXViT
   • Download: same as above

Dataset Access

Please obtain the dataset from the following link: MMPD‑Dataset

License

Code and data may be used freely for non‑commercial purposes; commercial inquiries should contact Sheng Jin (jinsheng13[at]foxmail[dot]com).

Citation

If you use our paper and code, please cite:

@inproceedings{zhang2024when,
  title={When Pedestrian Detection Meets Multi‑Modal Learning: Generalist Model and Benchmark Dataset},
  author={Zhang, Yi and Zeng, Wang and Jin, Sheng and Qian, Chen and Luo, Ping and Liu, Wentao},
  booktitle={European Conference on Computer Vision (ECCV)},
  year={2024},
  month={September}
}