Datasets:
video video 10 30 | scenario stringclasses 5
values | source stringclasses 1
value |
|---|---|---|
smoke | gopro_v2_qa | |
crossed_hands | gopro_v2_qa | |
object_occlusion | gopro_v2_qa | |
low_texture_reflective | gopro_v2_qa | |
fast_head_turn | gopro_v2_qa |
SIGNIQ Egocentric Hand-Pose Sample
Five short egocentric (head-mounted GoPro) manipulation clips for hand-pose / action-segment annotation. Web-optimized H.264 (1280×720, 30 fps, faststart) so they stream in-browser and support HTTP range requests — ready to load straight into a Label Studio Video timeline project.
| clip | scenario | length |
|---|---|---|
| smoke_10s | quick smoke clip | 10 s |
| crossed_hands_30s | hands crossing / bimanual | 30 s |
| object_occlusion_30s | hand occluded by object | 30 s |
| low_texture_reflective_30s | low-texture / reflective surface | 30 s |
| fast_head_turn_30s | fast head motion | 30 s |
Import into Label Studio / SIGNIQ Lab
videos/ holds the clips; label_studio_import.json is a ready task list pointing at the
hosted URLs. In a project using the Egocentric Hand Pose (Video Timeline) template:
Import → upload label_studio_import.json (or paste the URL list). Each task binds to the
config's single $video key.
Source: SIGNIQ Lab GoPro v2 QA capture set.
- Downloads last month
- 230