Video · Multimodal / AI & Machine Learning
Multimodal Human Manipulation Dataset | 6,000 Hours with RGB-D, Mocap, IMU & Hand Pose - Egocentric Data
Advertised by providerLicensing rights not independently verifiedPrice on request
Overview
Multimodal Human Manipulation Dataset | 6,000 Hours with RGB-D, Mocap, IMU & Hand Pose - Egocentric Data is a multimodal dataset. The provider states a volume of 6,000 hours. Typical applications include robotics & embodied AI. Relevant sectors: AI & Machine Learning, Education and Manufacturing & Industrial. The listing mentions timestamps. Pricing is not published and is provided on request.
Primary applications
What the listing offers
- Stated volume of 6,000 hours.
- Listing mentions timestamps.
The Multimodal Human Manipulation Dataset is a 6,000-hour sensor-rich collection of humans performing physical manipulation tasks with synchronized visual, motion, and sensor data. The dataset captures human demonstrations rather than robot executions, making it valuable for robotics and Physical AI teams that need richer physical-world supervision than video alone. Each recording includes first-person egocentric video, external camera views, RGB-D video, depth data, motion capture, IMU signals, hand pose, body pose, object labels, tool labels, timestamps, task phase labels, and completion states. This gives models a detailed view of how humans reach, grasp, transfer, fold, wipe, open, close, sort, place, and manipulate objects in real-world physical environments. Task coverage includes pick-and-place, object sorting, drawer opening and closing, folding, wiping, bin loading and unloading, shelf stocking, household manipulation, tool use, light assembly, package handling, object transfer, and other contact-rich workflows. The dataset supports physical reasoning, human demonstration learning, action recognition, motion modeling, manipulation research, embodied AI, and world model training for real-world object interaction. Technical specifications include 6,000 hours of synchronized multimodal human manipulation data, egocentric video, external camera views, RGB-D streams, depth maps, motion capture, IMU signals, hand pose, body pose, object labels, tool labels, action segments
Related datasets
More in Video & Physical AI →Video · MultimodalAdvertised
Egocentric Video Dataset | 100,000 hours of First-Person Videos for Physical AI, Robotics & Egocentric Vision
The Egocentric Residential Task Video Dataset is a 100,000-hour collection of high-quality first-person videos recorded in real residential environments across Brazil at…
- Language
- —
- Volume
- 100k h
- Domain
- AI & Machine Learning +2
- Use case
- Robotics & embodied AI
Price on request