dataset card
Browse files
README.md
CHANGED
|
@@ -1,17 +1,20 @@
|
|
| 1 |
# handTrackingSample
|
| 2 |
|
| 3 |
Egocentric footage of people working with their hands, cut to the sections a human picked as
|
| 4 |
-
hand-tracked, with synchronised IMU and
|
| 5 |
|
| 6 |
-
**
|
| 7 |
|
| 8 |
---
|
| 9 |
|
| 10 |
-
> **
|
| 11 |
-
>
|
| 12 |
-
>
|
| 13 |
-
>
|
| 14 |
-
>
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
|
| 17 |
## What each episode contains
|
|
@@ -22,7 +25,7 @@ hand-tracked, with synchronised IMU and per-frame hand pose.
|
|
| 22 |
right.mp4 the other eye of the stereo pair, same treatment
|
| 23 |
frames.csv frame_idx, ts_us <- the authoritative video clock
|
| 24 |
imu.csv ts_us, ax/ay/az (m/s2), gx/gy/gz (deg/s)
|
| 25 |
-
hands.jsonl per-frame hand
|
| 26 |
labels.json scene, task, job, environment, zone provenance
|
| 27 |
_stat.json processing record, including the exact anonymisation setting
|
| 28 |
```
|
|
@@ -47,25 +50,31 @@ Video and IMU stay locked to each other across them -- the sample-per-frame rati
|
|
| 47 |
integrating gyro on an assumed fixed `dt` will accumulate error. Gaps are listed per episode in
|
| 48 |
`_stat.json`.
|
| 49 |
|
| 50 |
-
**`hands.jsonl` is
|
| 51 |
-
|
|
|
|
| 52 |
|
| 53 |
---
|
| 54 |
|
| 55 |
## Anonymisation -- known limitation
|
| 56 |
|
| 57 |
-
Faces are detected with SCRFD-10G at a 960 letterbox, upright threshold 0.
|
| 58 |
-
four-rotation pass at 0.
|
|
|
|
| 59 |
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
|
|
|
|
|
|
| 64 |
|
| 65 |
-
**
|
| 66 |
contain faces some detector saw at all, so faces no model detected are absent from both the
|
| 67 |
-
numerator and the denominator. **Identifiable faces remain in this data
|
| 68 |
-
recognition, re-identification, or any purpose
|
|
|
|
|
|
|
| 69 |
|
| 70 |
## Lens geometry
|
| 71 |
|
|
@@ -77,7 +86,7 @@ raw geometry was judged the better trade. Per-camera intrinsics are available on
|
|
| 77 |
|
| 78 |
## Hand pose
|
| 79 |
|
| 80 |
-
21 3D joints per hand from **WiLoR** (ViT
|
| 81 |
MediaPipe's landmarker was evaluated against WiLoR and EgoHOS on this footage and was not good
|
| 82 |
enough -- it is the same model family that came last of eight detectors on our face benchmark, and
|
| 83 |
head-mounted video is its weak case.
|
|
@@ -85,27 +94,38 @@ head-mounted video is its weak case.
|
|
| 85 |
```json
|
| 86 |
{"frame_idx": 120,
|
| 87 |
"pose": [{"kp2d": [[x,y], ...21], "kp3d": [[x,y,z], ...21],
|
| 88 |
-
"is_right": true, "box": [x1,y1,x2,y2]
|
|
|
|
| 89 |
```
|
| 90 |
|
| 91 |
-
`kp2d` is in frame pixels; `box` is WiLoR's own hand detection, so detection and articulation
|
| 92 |
-
from the same model and always agree.
|
| 93 |
-
frame, not camera
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
|
| 95 |
-
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
|
| 98 |
## Scene and task spread
|
| 99 |
|
| 100 |
-
|
| 101 |
|
| 102 |
| category | recordings |
|
| 103 |
|---|---|
|
| 104 |
| Creative Workshops | 8 |
|
| 105 |
| Repair Services | 6 |
|
| 106 |
| Printing and Design | 5 |
|
| 107 |
-
| Food and Beverage |
|
| 108 |
-
| Retail and Consumer Goods |
|
| 109 |
| Cleaning and Sanitation | 4 |
|
| 110 |
| Laboratory / Scientific | 3 |
|
| 111 |
| Automotive and Transport | 3 |
|
|
@@ -115,21 +135,14 @@ runs separately and was not applied here; `hands.jsonl` carries hand detection a
|
|
| 115 |
| Industrial Manufacturing | 1 |
|
| 116 |
| Agriculture and Farming | 1 |
|
| 117 |
|
| 118 |
-
No category exceeds 8 of
|
| 119 |
set spans the full camera fleet rather than over-sampling whichever recordings happened to be
|
| 120 |
longest.
|
| 121 |
|
| 122 |
## How much of the footage carries a hand skeleton
|
| 123 |
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
- typical episodes land around **60-95%**
|
| 128 |
-
- bench work with both hands in view sits at the top of that range
|
| 129 |
-
- **overhead work is the weak case** -- an electrician working at ceiling height gave 1.8%, because
|
| 130 |
-
the camera spends most of the episode pointed at bare wall, and when the hands do appear the
|
| 131 |
-
wearer is reaching UP, so the arms enter from the top of frame. WiLoR is trained mostly on hands
|
| 132 |
-
entering from below, and an inverted approach is its hard case.
|
| 133 |
|
| 134 |
Check `hands.jsonl` before assuming a given episode is densely annotated.
|
| 135 |
|
|
@@ -137,4 +150,4 @@ Check `hands.jsonl` before assuming a given episode is densely annotated.
|
|
| 137 |
|
| 138 |
Every episode records its source recording, the exact source frame range, the curated zone it came
|
| 139 |
from, and the anonymisation setting used, in `_stat.json`. Cut and rendered from raw with
|
| 140 |
-
`generateSample/run_sample.py`.
|
|
|
|
| 1 |
# handTrackingSample
|
| 2 |
|
| 3 |
Egocentric footage of people working with their hands, cut to the sections a human picked as
|
| 4 |
+
hand-tracked, with synchronised IMU and hand pose on **every frame**.
|
| 5 |
|
| 6 |
+
**120 episodes · 8.4 hours · 44 recordings · 27 cameras · 1920x1200 · stereo**
|
| 7 |
|
| 8 |
---
|
| 9 |
|
| 10 |
+
> **19 Aug 2026 — RE-UPLOAD IN PROGRESS.** The 120 re-rendered episodes are being pushed now and
|
| 11 |
+
> will appear here over the next hour or so; the file tree below is incomplete until then.
|
| 12 |
+
>
|
| 13 |
+
> **Full re-release.** Every episode here was re-rendered from raw: no lens
|
| 14 |
+
> rectification (the earlier release had a visible circular seam on cameras whose undistortion map
|
| 15 |
+
> folds inside the frame), a stricter face-blur setting (see *Anonymisation*), no orientation flip
|
| 16 |
+
> guessing, and WiLoR hand pose on every frame rather than every sixth. The earlier 149-episode
|
| 17 |
+
> set has been withdrawn in full; this set replaces it.
|
| 18 |
|
| 19 |
|
| 20 |
## What each episode contains
|
|
|
|
| 25 |
right.mp4 the other eye of the stereo pair, same treatment
|
| 26 |
frames.csv frame_idx, ts_us <- the authoritative video clock
|
| 27 |
imu.csv ts_us, ax/ay/az (m/s2), gx/gy/gz (deg/s)
|
| 28 |
+
hands.jsonl per-frame hand boxes, 21-joint 2D/3D pose and MANO parameters
|
| 29 |
labels.json scene, task, job, environment, zone provenance
|
| 30 |
_stat.json processing record, including the exact anonymisation setting
|
| 31 |
```
|
|
|
|
| 50 |
integrating gyro on an assumed fixed `dt` will accumulate error. Gaps are listed per episode in
|
| 51 |
`_stat.json`.
|
| 52 |
|
| 53 |
+
**`hands.jsonl` is keyed by `frame_idx`.** There is one row per decoded frame, but index by
|
| 54 |
+
`frame_idx` rather than by position -- it is the same index `frames.csv` uses, and it is what ties
|
| 55 |
+
a row to its timestamp.
|
| 56 |
|
| 57 |
---
|
| 58 |
|
| 59 |
## Anonymisation -- known limitation
|
| 60 |
|
| 61 |
+
Faces are detected with SCRFD-10G at a 960 letterbox, upright threshold **0.80**, unioned with a
|
| 62 |
+
four-rotation pass at **0.80**, boxes over 10% of the frame additionally required to score 0.80,
|
| 63 |
+
and pixelated with an oval mask held for 6 frames after the last detection.
|
| 64 |
|
| 65 |
+
This is a deliberately **conservative** setting, chosen after the earlier release pixelated hands,
|
| 66 |
+
cookware and packaging that the detector scored as faces. On a hand-labelled benchmark of 31 faces
|
| 67 |
+
drawn from this same footage it catches **5 of 31 (16%)** with one false positive in 140 frames;
|
| 68 |
+
the earlier release's setting caught 68% with eight. The faces it does catch are the frontal,
|
| 69 |
+
close ones -- a bystander in profile or a face at an angle to the camera will usually score below
|
| 70 |
+
0.80 and pass through unblurred.
|
| 71 |
|
| 72 |
+
**16% is an upper bound on what was caught, not a floor on what leaked.** The benchmark can only
|
| 73 |
contain faces some detector saw at all, so faces no model detected are absent from both the
|
| 74 |
+
numerator and the denominator. **Identifiable faces remain in this data, more than in most
|
| 75 |
+
published egocentric sets.** Do not use it for face recognition, re-identification, or any purpose
|
| 76 |
+
requiring that individuals be unidentifiable. `_stat.json` records exactly what was applied, so
|
| 77 |
+
a stronger pass can be layered on top with your own detector.
|
| 78 |
|
| 79 |
## Lens geometry
|
| 80 |
|
|
|
|
| 86 |
|
| 87 |
## Hand pose
|
| 88 |
|
| 89 |
+
21 3D joints per hand from **WiLoR** (ViT backbone + MANO mesh head), on **every frame**.
|
| 90 |
MediaPipe's landmarker was evaluated against WiLoR and EgoHOS on this footage and was not good
|
| 91 |
enough -- it is the same model family that came last of eight detectors on our face benchmark, and
|
| 92 |
head-mounted video is its weak case.
|
|
|
|
| 94 |
```json
|
| 95 |
{"frame_idx": 120,
|
| 96 |
"pose": [{"kp2d": [[x,y], ...21], "kp3d": [[x,y,z], ...21],
|
| 97 |
+
"is_right": true, "box": [x1,y1,x2,y2],
|
| 98 |
+
"mano": {"global_orient": [3], "hand_pose": [45], "betas": [10], "cam_t": [3]}}]}
|
| 99 |
```
|
| 100 |
|
| 101 |
+
- `kp2d` is in frame pixels; `box` is WiLoR's own hand detection, so detection and articulation
|
| 102 |
+
come from the same model and always agree.
|
| 103 |
+
- `kp3d` is WiLoR's 3D joint prediction in its own metric frame (metres, hand-rooted), not camera
|
| 104 |
+
millimetres. `mano` carries the axis-angle global orientation, the 15 joint rotations, the shape
|
| 105 |
+
coefficients and the weak-perspective camera translation, so the full mesh can be rebuilt with a
|
| 106 |
+
standard MANO layer.
|
| 107 |
+
- Left hands are predicted by mirroring, as in WiLoR; `kp3d.x`, `global_orient` and `hand_pose`
|
| 108 |
+
are already mirrored back into the un-flipped frame.
|
| 109 |
+
- Frames with no detected hand carry `"pose": []`.
|
| 110 |
|
| 111 |
+
Pose was computed with a batched re-implementation of WiLoR-mini's inference loop (crop-local
|
| 112 |
+
anti-alias blur instead of a full-frame blur per hand). Against the stock pipeline on 102 hands it
|
| 113 |
+
gives identical detections and a median 1.0 px / max 7.6 px difference in `kp2d`, which is the
|
| 114 |
+
model's own sensitivity to sub-grey-level input rounding.
|
| 115 |
+
|
| 116 |
+
**Held-object boxes are not in this release.** `hands.jsonl` carries hand detection and pose only.
|
| 117 |
|
| 118 |
## Scene and task spread
|
| 119 |
|
| 120 |
+
120 episodes drawn from 44 recordings on 27 different cameras, across **13 top-level categories**:
|
| 121 |
|
| 122 |
| category | recordings |
|
| 123 |
|---|---|
|
| 124 |
| Creative Workshops | 8 |
|
| 125 |
| Repair Services | 6 |
|
| 126 |
| Printing and Design | 5 |
|
| 127 |
+
| Food and Beverage | 4 |
|
| 128 |
+
| Retail and Consumer Goods | 4 |
|
| 129 |
| Cleaning and Sanitation | 4 |
|
| 130 |
| Laboratory / Scientific | 3 |
|
| 131 |
| Automotive and Transport | 3 |
|
|
|
|
| 135 |
| Industrial Manufacturing | 1 |
|
| 136 |
| Agriculture and Farming | 1 |
|
| 137 |
|
| 138 |
+
No category exceeds 8 of 44 recordings. Episodes were selected round-robin across cameras, so the
|
| 139 |
set spans the full camera fleet rather than over-sampling whichever recordings happened to be
|
| 140 |
longest.
|
| 141 |
|
| 142 |
## How much of the footage carries a hand skeleton
|
| 143 |
|
| 144 |
+
The share of frames that carry at least one hand varies a lot by task, and is a property of the
|
| 145 |
+
work, not a failure of the pipeline: (per-episode figures will be filled in when the upload completes).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 146 |
|
| 147 |
Check `hands.jsonl` before assuming a given episode is densely annotated.
|
| 148 |
|
|
|
|
| 150 |
|
| 151 |
Every episode records its source recording, the exact source frame range, the curated zone it came
|
| 152 |
from, and the anonymisation setting used, in `_stat.json`. Cut and rendered from raw with
|
| 153 |
+
`generateSample/run_sample.py`; pose with `generateSample/wilor_pose.py`.
|