First-person dexterous hand activity across 50+ real-world environments.
Total Duration
1,000+ hours
Quality Check
100% Verified
A diverse collection of egocentric POV recordings capturing dexterous hand movements across 50+ distinct environments—kitchens, laundry rooms, gemstone factories, workshops, bakeries, and beyond. Filmed by 400+ contributors using the Pineye egocentric camera system, this dataset delivers authentic first-person perspectives of real-world manual tasks: cooking, cleaning, gemstone polishing, ironing, food prep, textile handling, and dozens more. Every clip is fully consented with complete provenance documentation and audit trails. Sourced and quality-verified by Luel for immediate use in embodied AI, robotics imitation learning, and action recognition research. Samples: https://huggingface.co/datasets/Luel-ai/Ego-Realm
PCM 16-bit
MP4 (H.264 codec)
30 fps
1920×1080 Full HD RGB
Rich metadata, action-narration, 3D hand tracing, and more upon request