Synchronized Ego-Exo Video for Cross-View Learning
Time-synchronized ego–exo video with calibration, task graphs and pose labels for cross-view learning and human-to-robot transfer. Scope a multi-view program with Shaip.
Dataset Specifications
| Target Buyer Teams | Robot Learning Lead; Multimodal Research Lead; World Model Team; Robotics Data Platform |
| Representative Tasks | Long-horizon household, warehouse, assembly and tool-use procedures |
| Capture / Modalities | First-person camera + 2–5 calibrated exocentric cameras; optional RGB-D, audio and IMU |
| Annotations / Metadata | Time synchronization, calibration, task graph, key steps, body/hand pose, object tracks and state changes |
| Delivery Formats | MP4/VRS + calibration files + JSON/Parquet metadata |
| Scale (Illustrative)* | 500–3,000 synchronized hours |
| Participants / Operators | 250–1,500 participants |
| Tasks / Scenarios | 30–150 tasks |
| Objects / Sites / Views | 3–6 views per session |
| Recommended Engagement Scope | 75–200 hours across 8–20 tasks with 3–5 synchronized cameras |
Extends Shaip's demonstrated egocentric, exocentric and multi-sensor collection operations.
View the related case study →Where This Dataset Is Used
This egocentric exocentric dataset is built for teams training cross-view learning, view-invariant perception and human-to-robot transfer models. Foundation model labs and robotics research teams use this ego exo video dataset as a synchronized multi-view dataset for cross-view correspondence, skill transfer from third-person observation, and procedural activity understanding. If you’re sourcing an embodied AI dataset or multi-camera task data with time-synchronized ego-exo capture, calibration, task graphs and pose labels, this dataset enables training that bridges first-person and third-person viewpoints.
Request a Sample of the Synchronized Egocentric–Exocentric Task Dataset
Tell us about the robot product or Physical AI system you’re building and we’ll respond with sample access, specifications and a proposal matched to your task.
"*" indicates required fields
