Egocentric Video: 100,000 Hours of First-Person Perspective Data

100,000 hours of video data collected from a first-person perspective, designed to support advanced computer vision and multimodal AI systems. The dataset was created by InfoBayAI and was last updated on June 3, 2026. It is intended for training models in video understanding and temporal reasoning.

Use Cases

• Train action recognition models based on the large-scale collection of egocentric video.
• Develop video understanding systems based on the temporal reasoning and scene context inherent in first-person footage.
• Fine-tune multimodal AI systems based on the visual and temporal data provided.
• Implement Reinforcement Learning with Human Feedback (RLHF) workflows based on the activity sequences present in the data.

Strengths

• Contains 100,000 hours of video data, providing substantial scale for model training.
• Explicitly designed for advanced computer vision, video understanding, and multimodal AI system development.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file formats are unknown, which may limit suitability assessment.
• Freshness should be verified as the last update timestamp is from the future (2026-06-03).

Provenance

Source

InfoBayAI

Freshness

Last updated 2026-06-03 06:30:52