SkillAgentSearch skills...

Sapiensid

No description available

Install / Use

/learn @mk-minchul/Sapiensid
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

SapiensID

SapiensID is an official repository for SapiensID: Foundation for Human Recognition CVPR 2025 (https://arxiv.org/pdf/2504.04708).

@inproceedings{kim2025sapiensid,
  title={SapiensID: Foundation for Human Recognition},
  author={Kim, Minchul and Ye, Dingqiang and Su, Yiyang and Liu, Feng and Liu, Xiaoming},
  booktitle={CVPR},
  year={2025}
}

There are two main components:

  • SapiensID: modeling
  • WebBody: dataset

SapiensID

The core modeling and evaluation framework that provides:

  • Person re-identification models and pipelines
  • Pretrained models included
  • Comprehensive evaluation on multiple datasets including:
    • PRCC
    • WebBody
    • Market-1501
    • MSMT17
    • LTCC
    • CelebReID
    • DeepChange
    • CCDA
    • CCVID
    • (we provide code for creating validation sets for each dataset but do not distribute the datasets)
  • Support for both single and multi-GPU evaluation
  • Feature extraction and metric computation
  • Model training and inference pipelines

Refer to tasks/sapiensID/README.md for more details.

WebBody

A dataset creation and management tool that:

  • Downloads and processes images from web sources
  • Handles image resizing and quality control
  • Manages data organization and storage
  • Supports parallel processing for large-scale dataset creation
  • Integrates with wandb for monitoring download progress

Refer to WebBody/README.md for more details.

Related Skills

View on GitHub
GitHub Stars24
CategoryDevelopment
Updated6d ago
Forks4

Languages

Python

Security Score

70/100

Audited on Mar 31, 2026

No findings