Egocentric video
Egocentric video is video captured from the perspective of a person or embodied participant carrying the camera. It is also called first-person video. Unlike an external camera observing a scene, it moves with the participant and emphasizes what is near their viewpoint during an activity.
Also known as: First-person video
How it works
Wearable capture can record interactions with tools, objects, and environments. Depending on the setup, recordings may be paired with timestamps, audio, annotations, or other sensors. Research datasets such as Ego4D illustrate the use of first-person recordings for understanding activity.
Interpreting the footage requires attention to camera position, motion, occlusion, and synchronization. The participant’s camera view is not identical to their attention, intention, or complete sensory experience.
Why it matters for licensing
Licensing first-person footage can involve rights in the recording and information about participants, bystanders, locations, and visible materials. A recipient also needs documentation of capture conditions and annotations. This definition describes the research concept and does not imply Rancher supplies wearable-camera datasets.
Example
Fictional example: A permitted recording of a training exercise shows a participant assembling a sample device from a head-mounted camera. Reviewers label visible actions while marking moments where the hands or tools leave the frame.
Limitations and misconceptions
Footage may omit key actions or contain motion blur and occlusion. It does not automatically provide robot commands, precise three-dimensional geometry, or the reasons behind a decision. Privacy and third-party material require review before disclosure.
Questions to ask
- What does the camera actually capture, and what remains outside the frame?
- Are timing, annotations, and any additional sensors aligned?
- Which participant, bystander, and content permissions apply?
Sources
- Ego4D — Dataset documentation · Accessed
- Ego-Exo4D — Paired egocentric and exocentric video dataset · Accessed