Human-Art: Multi-Scenario Human-Centric Data for Natural and Artificial Scenes

IDEA-Research released Human-Art in 2023 to provide a multi-scenario human-centric collection bridging natural and artificial visual domains. The data supports human pose estimation and image generation across diverse artistic styles and real-world settings.

Use Cases

• Human pose estimation using multi-scenario annotations
• Cross-domain image generation bridging natural and artificial scenes
• Benchmarking object detection on non-photorealistic human figures

Strengths

• Covers both natural and artificial (artistic) scenes
• Peer-reviewed research origin (CVPR 2023)
• Permissive Apache-2.0 license

Limitations

• Unknown total record count in provided metadata
• Potential domain shift between diverse artistic styles

Provenance

Source

IDEA-Research (CVPR 2023 paper)

Freshness

Last updated October 2023; reflects state-of-the-art research from CVPR 2023.