What is HumanLoop?
HumanLoop is a first-person dataset of everyday household activities captured in native home environments. Using head-mounted and body-worn cameras, we record unscripted interactions across the entire home — kitchen, laundry, living spaces — annotated with hand pose, segmentation, and action labels.
Example Activities
Data Quality Gates
Every recording is scored 0–100 by
pipeline/quality_check.py before it's
added to the dataset.
| Check | Threshold | Score Weight |
|---|---|---|
| Frame Rate | ≥ 29.9 FPS (GoPro standard) | +10 pts |
| Resolution | Height ≥ 1080p | +5 pts |
| Hand Detection Rate | ≥ 80% of sampled frames (MediaPipe Hands) | +25 / −20 pts |
| Both Hands Visible | ≥ 50% of sampled frames | +15 pts |
| Sharpness (Blur Score) | Laplacian variance ≥ 500 | +20 / −25 pts |
| Brightness | Avg pixel value between 80–180 | +15 pts |
| Audio Narration | Speech detected via Whisper (no_speech_prob < 0.5) | +10 / −15 pts |
HomeHands-50
Egocentric footage of everyday household activities captured from a first-person wearable camera. The foundation for all subsequent HomeHands derived datasets.
21-point hand skeleton tracked per frame using MediaPipe HandLandmarker. Wrist, knuckles, and fingertips detected on both hands.
Per-frame instance masks for hands, objects, and surfaces using a fine-tuned segmentation model across all activity clips.
Live-narrated egocentric recordings with burned-in captions. The person performing each task describes their own actions in real time.
HomeDepth
Per-frame depth maps estimated from monocular RGB frames, capturing spatial layout and object distances for each household activity clip.
Want to Contribute Recordings or Pipeline Tools?
We accept new egocentric recordings and pipeline tooling contributions (preprocessing, validation, annotation scripts). Submissions are scored by the same quality-check pipeline used internally — see the tab.
Recording Session Quality
Scored by pipeline/quality_check.py
Sample sessions ranked by automated quality score. Replace with live results from
assets/processed/quality/*.json.
| Rank | Clip / Session | Hand Detection | Blur Score | Verdict | Quality Score |
|---|
Contribute to HumanLoop
This sandbox is an interactive demo of the quality-check pipeline. For real contributions, run
pipeline/quality_check.py locally or
reach out via the contact card.