CUSTOMVIDEODEXTEROUS TASKS
Hand and Object Interaction Clips
Hand-object interaction clips with action verbs, object categories, camera angle metadata, contact-event markers, object-state labels, and QA fields for robotic manipulation evaluation.
CUSTOMVIDEOFIRST-PERSON
First-Person Activity Clips
First-person video collection format for action understanding, hand-object reasoning, embodied AI evaluation, robotics perception, and vision-language grounding.
First-Person Activity Clips are built for teams evaluating models that need to understand activity from the actor’s point of view. The format captures physical task flow, hand movement, object interaction, environmental context, step order, capture perspective, and completion patterns.
The format is designed for embodied systems that need video sequences with action context rather than isolated frames.
Structured collection
First-Person Activity Clips use defined physical tasks and capture instructions.
Metadata depth
The format supports alignment between video timestamps, step labels, narration transcript where scoped, action labels, and task…
Delivery-ready package
Includes video quality, task visibility, hands-visible reviews, plus dataset manifest, CSV catalog, consent status metadata, QA…
Request sample access
Share your target task types, capture setup, annotation needs, and evaluation goals. We’ll reply with a technical scope for the collection.
Key highlights
- First-person task video with capture perspective, environment metadata, and task labels.
- Hands-visible review for tasks where hand-object reasoning is central.
- Timestamped step metadata and optional narration alignment.
Dataset contents
Video capture
First-person video clips · Task labels · Step or phase metadata · Environment type metadata · Camera perspective metadata
Steps, transcripts, and labels
Device model metadata · Resolution and fps · Duration metadata · Hands-visible review fields · Lighting condition metadata
Capture and task metadata
Optional narration track · Optional action labels · Optional IMU data where available
Technical specifications
Use cases
- Embodied AI evaluation
- Action recognition
- Hand-object reasoning
- Robotics perception research
- Vision-language grounding
- Procedural understanding
- Egocentric world modeling
- Imitation learning evaluation
Dataset workflow
1Collection
First-Person Activity Clips use defined physical tasks and capture instructions. Contributors record approved activities from a first-person perspective while preserving task flow, hand interaction, environment context, and completion sequence.
2Review
Video quality review · Task visibility review · Hands-visible review · Step alignment review · Environment metadata review · Consent status review · Delivery approval
3Delivery
MP4 video clips · Task metadata CSV · Step metadata JSON where scoped · Annotation files where scoped · Capture metadata CSV · Dataset manifest JSON · CSV file catalog · QA summary · Consent status metadata · Delivery notes
Quality assurance
Related dataset formats
CUSTOMVIDEOINSTRUCTIONAL
Procedural Demonstration Video Sets
Step-structured procedural videos with task labels, timestamped step markers, narration alignment, completion status, and metadata for instruction-following and imitation learning.
CUSTOMVIDEOENVIRONMENT DIVERSITY
Multi-Environment Activity Clips
Activity video sets captured across varied real-world environments, with task labels, environment metadata, camera position fields, lighting conditions, duration, and QA status for activity recognition and embodied AI evaluation.
CUSTOMMULTIMODALPAIRED DATA
Multimodal Paired Capture Sets
Paired multimodal datasets linking video, audio, screen, text, or image-derived records through shared file IDs, cross-modal manifests, metadata schemas, and QA fields.
CUSTOMAUDIOACOUSTIC SCENES
Acoustic Environment Sets
Ambient audio sets with scene labels, acoustic context metadata, speech screening, device fields, noise-level estimates, and QA status for acoustic scene and robustness evaluation.
CUSTOMANY MODALITYBUYER-SCOPED
Buyer-Scoped Custom Collection
Fully scoped custom data collection with buyer-defined modality, schema, metadata fields, annotation layers, QA criteria, manifest structure, delivery package, and collection rules.