CHIRLA: 1M Identity-Annotated Bounding Boxes for Long-Term Re-ID

CHIRLA provides approximately 1 million identity-annotated bounding boxes across 596,000 frames for long-term person re-identification and tracking. Captured by bdager over a 7-month period, the data utilizes 7 cameras within connected indoor environments to track 22 unique identities.

Use Cases

• Training re-identification models to recognize the 22 identities despite clothing changes
• Developing multi-camera tracking systems using the 7-camera indoor network
• Analyzing long-term appearance variation in identity-annotated bounding boxes

Strengths

• ~1M identity-annotated bounding boxes
• 7-month longitudinal temporal span
• Multi-camera coverage across 7 connected indoor views

Limitations

• Small identity diversity with only 22 unique individuals
• Limited to indoor environmental conditions
• High frame-to-frame redundancy across 596,000 frames

Provenance

Source

bdager (Arxiv 2502.06681)

Collection Method

Multi-camera sensor capture and identity annotation

Time Range

7 months (ending 2025)

Freshness

Last updated October 2025

Geography

United States