Publish three owned egocentric RGB-D kitchen captures under CC BY 4.0

LICENSE

@@ -0,0 +1,7 @@
+Creative Commons Attribution 4.0 International (CC BY 4.0)
+
+Copyright 2026 Diffraction. Licensed under https://creativecommons.org/licenses/by/4.0/
+Legal terms: https://creativecommons.org/licenses/by/4.0/legalcode
+Attribution: Diffraction Egocentric Kitchen Capture Sample, dataset revision and source URL.
+
+Dataset URL: https://huggingface.co/datasets/diffracting/egocentric-kitchen-sample

QUALITY.md

@@ -0,0 +1,29 @@
+# Sample quality and validation
+
+This release contains three owned kitchen recordings, totaling 167.85 seconds and 10,070 displayed RGB frames at 1920 x 1440. Native depth/confidence images are 256 x 192. All recordings remain in one conservative session group; they are a technical sample, not a train/test benchmark.
+
+## Sensor integrity
+
+Every displayed RGB frame has a corresponding source frame ID, depth/confidence pair, camera pose and per-frame intrinsic calibration. Each source also has one discarded pre-roll video packet; its original sensor record is retained with a null RGB index. This explains the difference below. The original clocks, video bytes and sensor payloads are preserved. Camera pose is a device VIO estimate, not independent ground truth. Raw IMU acceleration units remain unverified for the capture app version.
+
+| Recording | Displayed RGB | Sensor records | Unknown hand label rows | Timestamps with >2 temporal hands |
+|---|---:|---:|---:|---:|
+| onion-preparation | 4,543 | 4,544 | 69.7% | 49/1136 |
+| pepper-slicing | 3,465 | 3,466 | 95.6% | 26/866 |
+| pepper-dicing | 2,062 | 2,063 | 95.2% | 0/515 |
+
+## Optional annotation quality
+
+The RGB-D capture is the primary data. RTMPose/RTMW hand estimates, OWLv2 object boxes, temporal tracks, depth-lifted wrists and contact estimates are experimental auxiliary channels. Raw detections are capped at two hands after duplicate suppression; temporal gap filling can retain short extra tracks. The counts above expose this behavior. Do not treat track IDs as verified anatomical identity. Conflicting left/right predictions are unknown.
+
+Sampled visual inspection of three evenly spaced points in every final overlay found useful detections alongside occlusion errors, loose object boxes and ambiguous hand identity. No independent keypoint, contact or tracking accuracy benchmark has been conducted. Per-frame scores are not calibrated error probabilities. Missing 3D hand outputs remain null; palm orientation is frequently unavailable. Annotated overlays are provided under `review/` for inspection and do not alter native training pixels.
+
+## Validation evidence
+
+`validation.json` records actual consumer checks: a standalone reader in an isolated environment decoded five positions in every video, including the first and last; it loaded the corresponding depth/confidence/calibration/pose, checked geometry and missing-value semantics, loaded the three video rows with Hugging Face Datasets, and verified packaged SHA-256 hashes. These checks establish packaging integrity, not annotation accuracy or robot transfer. `source_preservation.json` compares released video and sensor payloads with the three approved source archives.
+
+An automated full-video face scan returned zero face detections for all three recordings. Sampled visual inspection found kitchen work surfaces and the wearer's hands. This is not an independent human privacy audit. The owner authorized these recordings and selected CC BY 4.0.
+
+## Next evidence to collect with a lab
+
+A pilot should specify activities, participant/session diversity, device and mount metadata, environment conditions and the acceptance criteria before capture. A small independently labeled subset can quantify hand visibility, keypoint error, object-track stability and action boundaries. A downstream ablation can then test whether native RGB-D and optional annotations improve the lab's target task. These results are not claimed by this sample.

README.md

@@ -0,0 +1,82 @@
+---
+license: cc-by-4.0
+pretty_name: Diffraction Egocentric Kitchen Capture Sample
+configs:
+- config_name: default
+  data_files:
+  - split: sample
+    path: data/episodes.parquet
+language:
+- en
+tags:
+- video
+- robotics
+- egocentric
+- rgbd
+- physical-ai
+- human-demonstration
+- depth-estimation
+size_categories:
+- n<1K
+---
+# Diffraction Egocentric Kitchen Capture Sample
+
+A small, inspectable sample of human kitchen manipulation captured with Stray Scanner on a LiDAR-equipped iPhone: native RGB, metric depth and confidence, per-frame camera calibration, device odometry, raw device IMU, and explicitly estimated hand/object annotations.
+
+**Human observation sample.** License: cc-by-4.0. This sample contains 3 recordings totaling 167.85 seconds. It is an observation dataset for evaluating human-video pretraining, spatial perception, action understanding and motion-prediction workflows. It supplies no measured robot commands, force or tactile data, and demonstrates no robot-policy performance gain.
+
+| Capture | Seconds | Native RGB frames | Instruction |
+|---|---:|---:|---|
+| onion-preparation | 75.73 | 4,543 | Prepare and peel the onion on the cutting board using hands and a kitchen knife. |
+| pepper-slicing | 57.75 | 3,465 | Prepare and slice the bell pepper on the cutting board using a kitchen knife. |
+| pepper-dicing | 34.37 | 2,062 | Dice the bell pepper strips on the cutting board using a kitchen knife. |
+
+![Native RGB, metric depth and confidence](media/sample-overview.jpg)
+
+## Download and try one episode
+
+```bash
+pip install huggingface_hub
+hf download diffracting/egocentric-kitchen-sample --repo-type dataset --local-dir diffraction-sample
+cd diffraction-sample
+pip install -r requirements.txt
+python observation_reader.py .
+```
+
+The download includes native RGB-D recordings and a portable reader. The Hugging Face dataset viewer uses smaller video previews; use `videos/` for native RGB training pixels.
+
+```python
+from observation_reader import ObservationDataset, camera_points
+sample = ObservationDataset('.')
+key = next(iter(sample.episodes))
+frame = sample.frame(key, 100)
+print(frame['rgb'].shape, frame['depth_m'].shape, frame['K_rgb'])
+points = camera_points(frame)
+hands = sample.annotations_near(key, 'hand_pose', frame['timestamp_s'], max_age_s=0.05)
+```
+
+Native videos are under `videos/`; smaller browser previews and episode metadata under `previews/`. Depth/confidence and original sensor CSVs are in `sensors/`. Annotation tables are standalone Parquet. [SCHEMA.md](SCHEMA.md) defines units, frames, clock mapping, confidence and missing-data behavior. `checksums.sha256` covers every packaged artifact.
+
+## Collection and scope
+
+The owner confirmed these Stray Scanner recordings are owned and may be used to prepare a training-data sample. This small collection shows closely related kitchen tasks; participant count, device model/app version and independent session identity were not supplied. All recordings belong to a conservative common session group. There is no claimed train/test benchmark or cross-person generalization.
+
+The pipeline preserves native video bytes and joins camera/depth records by source frame ID, retaining both original clocks. Per-frame intrinsics are used for wrist lifting. Raw and temporally processed keypoints are separate. Instructions were prepared from video inspection; detailed action boundaries, success/failure and ground-truth contact annotations are not provided.
+
+## Quality and limitations
+
+Per-episode `quality/` reports expose channel coverage, null counts, provenance and sensor correspondence. These are engineering checks, not labeled accuracy benchmarks. ARKit camera pose is a device VIO estimate. Hand pose, palm orientation, object tracking and contact remain estimates; review their overlays and confidence/missingness before training. The low-resolution depth sensor can miss thin fingers, tool edges and occlusion boundaries. Fast cutting, bimanual occlusion and deformable vegetables are difficult cases.
+
+Raw IMU acceleration units need capture-version verification; no silent unit conversion is applied. Device IMU is not wrist-motion sensing. See SCHEMA.md. An independent labeled evaluation and robot transfer experiment are future work, not measured results of this sample.
+
+## Reproducibility and professional capture
+
+The accompanying processing record identifies the source code and environment used. `QUALITY.md` summarizes observed coverage and limitations; `validation.json` records the portable-reader and file-integrity checks. Run `python validate_release.py .` after installing `datasets` alongside the reader requirements to repeat these checks. This sample is intended to start technical evaluation of task-specific professional capture: agree on target activities, environments, collection protocol, annotation requirements and measurable acceptance criteria before scaling a collection.
+
+## References
+
+[Stray Scanner source format](https://docs.strayrobots.io/apps/scanner/format.html), [Hugging Face video datasets](https://huggingface.co/docs/hub/datasets-video).
+
+## Attribution
+
+Please attribute **Diffraction, Egocentric Kitchen Capture Sample (2026)** and include the [dataset URL](https://huggingface.co/datasets/diffracting/egocentric-kitchen-sample) and the revision you used. The dataset is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).

SCHEMA.md

@@ -0,0 +1,15 @@
+# Observation schema v1.0.0
+
+`dataset.json` indexes complete recordings in the `sample` subset. Reader RGB indices are zero-based and refer to displayed video frames. `source_frame_index` identifies the original Stray sensor record and can differ from `rgb_frame_index`: these recordings contain a discarded pre-roll packet. Both mappings are retained. Browser previews are resized review copies.
+
+- `frame_mapping`: source_frame_index, native video_pts_s, device sensor_timestamp_s, independently zeroed sensor_t_s, and canonical t_s (video PTS minus the first displayed frame PTS), rgb_decodable and rgb_frame_index. Discarded pre-roll records have null RGB indices and may have negative t_s; they remain in the raw sensor tables but are not training RGB samples. Mapping is by explicit source frame ID, not nearest independently zeroed clock. Clock agreement is not an independent physical synchronization measurement.
+- `camera_intrinsics`: per-frame fx/fy/cx/cy in native RGB pixels, width/height, calibration_source. Downsample focal/principal-point values by depth/RGB resolution ratio for depth unprojection.
+- `arkit_poses`: t_s, tx/ty/tz in metres and quaternion qx/qy/qz/qw. Camera-to-world in ARKit gravity-aligned world, Y up. Device visual-inertial estimates, not motion-capture truth. OpenCV camera axes are x right, y down, z forward; flip y/z before applying ARKit camera-to-world.
+- `sensors/*.zip`: original depth PNGs in millimetres and confidence PNGs (0 low, 1 medium, 2 high); CSV sensor payloads preserved byte-for-byte. Reader returns float32 metres and masks confidence <1 and zero depth. Preserve confidence as ordinal levels, not probabilities.
+- `device_imu`: raw a_x/a_y/a_z and alpha_x/alpha_y/alpha_z (rad/s per source documentation), device timestamps and mapped video t_s. Acceleration units are native/unverified: published docs say m/s^2 but observed magnitudes suggest g-scale logging. No conversion, gravity removal or device-to-camera rotation is assumed. Out-of-coverage t_s is null with clock_mapping_valid=false. This is a phone sensor, not a wrist sensor.
+- `hand_pose_raw` and `hand_pose`: 21 estimated 2D keypoints per hand, normalized x/y, track ID, handedness, scores. z/wx/wy/wz are null for the 2D detector. Temporal outputs carry detected/smoothed/interpolated/extrapolated provenance. Detection scores are not calibrated error probabilities.
+- `hand_world_trajectory`: depth-anchored estimated wrist position; optional estimated palm-plane xyzw orientation, orient_valid. Orientation null when unavailable. Metric coordinates do not imply measured anatomical accuracy. The 2D detector and low-resolution depth are error sources. The single-wearer recipe suppresses duplicate candidates, caps raw detections at two hands (temporal gap filling can retain additional short tracks), and marks conflicting model handedness unknown; original candidate selection counts are disclosed in quality metadata.
+- `tracks*`, `contact*`: experimental object boxes and hand/object proximity estimates. Cutting deforms objects; instance identity and true physical contact are not ground truth. Contact has no force/tactile supervision.
+- `quality/*.json`: completeness, coverage, missingness and processing metadata. No aggregate accuracy score. No robot action, reward or terminal-success field is fabricated.
+
+Signals retain their native sampling rates. A 60 fps RGB stream does not mean 60 independent hand estimates per second. Use bounded joins (`annotations_near`) and preserve absent values. Never forward-fill across long gaps.

annotations/onion-preparation/arkit_poses.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f87250f64cf7cc2f3b7721be16e6bdca83201013f6f763b4a6a93ddac6ec5018
+size 242601

annotations/onion-preparation/camera_intrinsics.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cbc507375a40a282655bb866cfbbe519dc6b5d686985e2a0ad02f80932d96e91
+size 87913

annotations/onion-preparation/contact.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99203e53d7b1cbb777a2a6c2c1aefe4747b40dc11bd02b8daf3fd84ad50a6a53
+size 55798

annotations/onion-preparation/contact_events.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87e3681cc3c28e722d166236038269e22b6a915dc684078e1ff83b81e14687b2
+size 6013

annotations/onion-preparation/device_imu.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78e1716c4e6486bae6a180be70afe12a6a0e0d24afa234d832ca32ddaede46d5
+size 316676

annotations/onion-preparation/frame_mapping.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:769db69a549db519a3cb499bbd5abbf12a0bca881e68ccec50e4d51651ed3cab
+size 94851

annotations/onion-preparation/hand_pose.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ecb5b271036cee5fd957922497e5b40a59e20fc66f97283dd6cf2298d8f5f830
+size 1055106

annotations/onion-preparation/hand_pose_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:036e8a220f1bf1facdcd40ebe28002958f795b7bf54fc0f995a08bf99fc26dda
+size 682283

annotations/onion-preparation/hand_world_trajectory.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ec866a09ab090f29f029ff8dae3c90a4e73eeb3fd8b320ef1a5f02b8b3f4665
+size 98589

annotations/onion-preparation/quality.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55b44176251829972f6fde9ab5c01d15010f8ccc538b778c9132bd6f22e56d07
+size 2497

annotations/onion-preparation/tracks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:964b03a16734fa43fa99c4e09b7c37ffa373d9ddb4733f59991c9a3d008d22ba
+size 53132

annotations/onion-preparation/tracks_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e289b8f821e4dc31569fc456cc6aea9b410784a1ff6e47cdc9cefa32f8773df6
+size 31445

annotations/pepper-dicing/arkit_poses.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9cfa9a6519ba94384cd4bb0fd1bcf8b90db388cde8cdb56c56418e6e1b00370
+size 113172

annotations/pepper-dicing/camera_intrinsics.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e672156868b3cf3a7869f0a37018360fe213ff2ad14d27273a7d3fac2eed2c3
+size 48278

annotations/pepper-dicing/contact.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d4c7f6997c9509223f95159567b52316d24a5171bee03e6e4867200adde5deea
+size 30905

annotations/pepper-dicing/contact_events.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05f36ba0a04941573eed0e67d0217c12f6e1a755e7c1d8f4862985a46b9576a4
+size 5177

annotations/pepper-dicing/device_imu.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7379566c8ce17c59e72109e57fe4c786a4d6a4557f1613b16f5026eb540538e5
+size 145155

annotations/pepper-dicing/frame_mapping.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5791bd68a3bd93ff83720cf48cf4d3e20d19830c2f8dcac25d397ffc8ceb4ac8
+size 44371

annotations/pepper-dicing/hand_pose.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3706ecb644961b1a7f8603dc03539266078282505f1acb0b9270b17120ccfdff
+size 464051

annotations/pepper-dicing/hand_pose_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:449be4ef4613817dc415768734bd345d62da49a44febcd646e61c7b566900079
+size 304439

annotations/pepper-dicing/hand_world_trajectory.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c15fa2ea991449f20e167ef928db953556fe4d8138d0ea422da6c0f2df5f9858
+size 42773

annotations/pepper-dicing/quality.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bab954c9e617c3c0f227bb5c767c701234b4b3e72bc0b4e102088497c78ab9f
+size 1750

annotations/pepper-dicing/tracks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b18f242e41e00a368119d56fbd746ad99dc2bc6de03954666baddc8aaba83a7
+size 25486

annotations/pepper-dicing/tracks_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e9a4e3f4c88837a6d1e28d3bc689c4823eb94556d79354c5a4e89549f3b12069
+size 15936

annotations/pepper-slicing/arkit_poses.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9cfda82ca4875e92c190a56e6ec4903568914ae8b9f556b660bd55a45588967
+size 187998

annotations/pepper-slicing/camera_intrinsics.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8891a8e13390b49455eaac3a6d0b051aca37690ac4b69be2f6f1d61513c6d627
+size 71048

annotations/pepper-slicing/contact.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8feadeeef6993735ce6ab1bcf4c41d0a395bc0d461ef6d8b2d0acea1bd274730
+size 43849

annotations/pepper-slicing/contact_events.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56a322addcc9bced3eb2691e1a30aaf4df3dc673196370a60eac6a11f891a919
+size 5985

annotations/pepper-slicing/device_imu.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:831b63695c5f0e23d6670bd32a3b1f0308a8823b222ccf8af244147cb07422ea
+size 242688

annotations/pepper-slicing/frame_mapping.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4aa93f1f76a07ed2f082358f5f109e364d483620ff6173906d42715f68cd923a
+size 73079

annotations/pepper-slicing/hand_pose.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5e2eb84b1526c2072e873369dbf510c3329500a8c28764b19dedca38df9a119
+size 816535

annotations/pepper-slicing/hand_pose_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8a59ba9662de33f55f7545e45cfc8ce90b755d30bd99da15beb757b8d67d2b4
+size 548113

annotations/pepper-slicing/hand_world_trajectory.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e374630e2eba6fa16819a441ff4a5f8832f1cfb0a059216ce90bcffb64351415
+size 70266

annotations/pepper-slicing/quality.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e511784770d9db114f6575b6dc900af1ff2bcd19f70ed4cb6b07b775fca5ebab
+size 2201

annotations/pepper-slicing/tracks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e8b9833e67315f243e279bae346558e229dfddd6cc7e7e3882e54946434c50f
+size 38818

annotations/pepper-slicing/tracks_raw.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf1980e5456c0158a741dc6c7df4246c67a9c87e00e86564144bc9d38ca8593f
+size 22865

checksums.sha256

@@ -0,0 +1,68 @@
+f87250f64cf7cc2f3b7721be16e6bdca83201013f6f763b4a6a93ddac6ec5018  annotations/onion-preparation/arkit_poses.parquet
+cbc507375a40a282655bb866cfbbe519dc6b5d686985e2a0ad02f80932d96e91  annotations/onion-preparation/camera_intrinsics.parquet
+99203e53d7b1cbb777a2a6c2c1aefe4747b40dc11bd02b8daf3fd84ad50a6a53  annotations/onion-preparation/contact.parquet
+87e3681cc3c28e722d166236038269e22b6a915dc684078e1ff83b81e14687b2  annotations/onion-preparation/contact_events.parquet
+78e1716c4e6486bae6a180be70afe12a6a0e0d24afa234d832ca32ddaede46d5  annotations/onion-preparation/device_imu.parquet
+769db69a549db519a3cb499bbd5abbf12a0bca881e68ccec50e4d51651ed3cab  annotations/onion-preparation/frame_mapping.parquet
+ecb5b271036cee5fd957922497e5b40a59e20fc66f97283dd6cf2298d8f5f830  annotations/onion-preparation/hand_pose.parquet
+036e8a220f1bf1facdcd40ebe28002958f795b7bf54fc0f995a08bf99fc26dda  annotations/onion-preparation/hand_pose_raw.parquet
+1ec866a09ab090f29f029ff8dae3c90a4e73eeb3fd8b320ef1a5f02b8b3f4665  annotations/onion-preparation/hand_world_trajectory.parquet
+55b44176251829972f6fde9ab5c01d15010f8ccc538b778c9132bd6f22e56d07  annotations/onion-preparation/quality.parquet
+964b03a16734fa43fa99c4e09b7c37ffa373d9ddb4733f59991c9a3d008d22ba  annotations/onion-preparation/tracks.parquet
+e289b8f821e4dc31569fc456cc6aea9b410784a1ff6e47cdc9cefa32f8773df6  annotations/onion-preparation/tracks_raw.parquet
+b9cfa9a6519ba94384cd4bb0fd1bcf8b90db388cde8cdb56c56418e6e1b00370  annotations/pepper-dicing/arkit_poses.parquet
+7e672156868b3cf3a7869f0a37018360fe213ff2ad14d27273a7d3fac2eed2c3  annotations/pepper-dicing/camera_intrinsics.parquet
+d4c7f6997c9509223f95159567b52316d24a5171bee03e6e4867200adde5deea  annotations/pepper-dicing/contact.parquet
+05f36ba0a04941573eed0e67d0217c12f6e1a755e7c1d8f4862985a46b9576a4  annotations/pepper-dicing/contact_events.parquet
+7379566c8ce17c59e72109e57fe4c786a4d6a4557f1613b16f5026eb540538e5  annotations/pepper-dicing/device_imu.parquet
+5791bd68a3bd93ff83720cf48cf4d3e20d19830c2f8dcac25d397ffc8ceb4ac8  annotations/pepper-dicing/frame_mapping.parquet
+3706ecb644961b1a7f8603dc03539266078282505f1acb0b9270b17120ccfdff  annotations/pepper-dicing/hand_pose.parquet
+449be4ef4613817dc415768734bd345d62da49a44febcd646e61c7b566900079  annotations/pepper-dicing/hand_pose_raw.parquet
+c15fa2ea991449f20e167ef928db953556fe4d8138d0ea422da6c0f2df5f9858  annotations/pepper-dicing/hand_world_trajectory.parquet
+9bab954c9e617c3c0f227bb5c767c701234b4b3e72bc0b4e102088497c78ab9f  annotations/pepper-dicing/quality.parquet
+9b18f242e41e00a368119d56fbd746ad99dc2bc6de03954666baddc8aaba83a7  annotations/pepper-dicing/tracks.parquet
+e9a4e3f4c88837a6d1e28d3bc689c4823eb94556d79354c5a4e89549f3b12069  annotations/pepper-dicing/tracks_raw.parquet
+b9cfda82ca4875e92c190a56e6ec4903568914ae8b9f556b660bd55a45588967  annotations/pepper-slicing/arkit_poses.parquet
+8891a8e13390b49455eaac3a6d0b051aca37690ac4b69be2f6f1d61513c6d627  annotations/pepper-slicing/camera_intrinsics.parquet
+8feadeeef6993735ce6ab1bcf4c41d0a395bc0d461ef6d8b2d0acea1bd274730  annotations/pepper-slicing/contact.parquet
+56a322addcc9bced3eb2691e1a30aaf4df3dc673196370a60eac6a11f891a919  annotations/pepper-slicing/contact_events.parquet
+831b63695c5f0e23d6670bd32a3b1f0308a8823b222ccf8af244147cb07422ea  annotations/pepper-slicing/device_imu.parquet
+4aa93f1f76a07ed2f082358f5f109e364d483620ff6173906d42715f68cd923a  annotations/pepper-slicing/frame_mapping.parquet
+d5e2eb84b1526c2072e873369dbf510c3329500a8c28764b19dedca38df9a119  annotations/pepper-slicing/hand_pose.parquet
+b8a59ba9662de33f55f7545e45cfc8ce90b755d30bd99da15beb757b8d67d2b4  annotations/pepper-slicing/hand_pose_raw.parquet
+e374630e2eba6fa16819a441ff4a5f8832f1cfb0a059216ce90bcffb64351415  annotations/pepper-slicing/hand_world_trajectory.parquet
+e511784770d9db114f6575b6dc900af1ff2bcd19f70ed4cb6b07b775fca5ebab  annotations/pepper-slicing/quality.parquet
+5e8b9833e67315f243e279bae346558e229dfddd6cc7e7e3882e54946434c50f  annotations/pepper-slicing/tracks.parquet
+cf1980e5456c0158a741dc6c7df4246c67a9c87e00e86564144bc9d38ca8593f  annotations/pepper-slicing/tracks_raw.parquet
+8cd1120d8395e43bd67a9bff6d570ef10323260f2e88f0e1641db67c8b9da99a  data/episodes.parquet
+eb673f030ae68ea2d4385e03f7db8b206e81028f24286da2a605a8e4d5c82fbe  dataset.json
+d2792efd3746fe2646e0a1813567e86fdcf4f28d85e60d6eed56d6f74bdb97ab  episodes/onion-preparation.json
+9d37b0a49a4f17841454ccc3419e3af1528e0ea10d5676837b4d90a75f0a0aad  episodes/pepper-dicing.json
+e4b1612b961a2ae031abb2fb315a2aba1bc3194c8c809e77560ac3e1e8e6d597  episodes/pepper-slicing.json
+d23d8443b516f623d4d726acd01b2b6a0f980d1892955322630ee8a9d6a2e836  LICENSE
+28573b0781f45e2ddb8ec9993fcc3bce127c57986c346dbb38723c034eeaf303  media/sample-overview.jpg
+a3cd51e1920831acde188b61081071e46fbebb3a68e07f6dfc094342188ea2f0  observation_reader.py
+a0a297083c733f3f7593475890e7e78c00c2a819c22484351a8305f5ef0bbd14  previews/metadata.jsonl
+37179bfaad964492247c48ef1418bdee941a62621d84796e4ca23144e5d195b2  previews/onion-preparation.mp4
+73bbb6078f833b536f93af57f2c2f825d43f73754a2ac950dde928b2bb6a5799  previews/pepper-dicing.mp4
+26abb7272170dc4c390882dc7330040bef0326f955690aaeabcdcb82cbf38c60  previews/pepper-slicing.mp4
+59dbee03309cfeda9f347f1a360e417b23ccb59870333eba787058d6f4ed6849  processing_record.json
+a348071bab435eb8be042ac9c7dcabc64fc1b215d7113c7d58dd91e85a3c4fd8  quality/onion-preparation.json
+a7b0c1c53883a774aeebb7422822d627c3bad25e05dca7cc7deee6269d61f5d9  quality/pepper-dicing.json
+4b4edd0120bcf67ff41dab1818d2902a1d2c6399872d7d149fa1936b3d726922  quality/pepper-slicing.json
+78ccb58981f0db46f83a0dfdc318883c75bc8938ca9be800aedce493504f2eb8  QUALITY.md
+2722f3f92272d5986dd0e7fe12483d0acf3ac5be3c6603defcb3d3e43d2ffd33  README.md
+4c2c8605ecfe1c00fc08c4bd686aef27d8bfd5ea790c1b216640c8a9e90db007  requirements.txt
+9ee9d3c75da721a39c65cefe2004231564bc9298c6a26b8dae623fdc8108e262  review/onion-preparation-overlay.mp4
+aad594507f78e2667c9092bc756a1ab4fba4fe949f5bca8eceb52e93695d14bb  review/pepper-dicing-overlay.mp4
+caf98485a6e87585fe9cdfbcfd8ec2505d2cf0c25903be492df6c2a9c9024a90  review/pepper-slicing-overlay.mp4
+535b53ca25abd3ea720ae80e7a3f92d476b17499161777a7560a7e84d5247234  SCHEMA.md
+f3257b74551aed7ed90adffc7d2462a11234b984ad9d2b393c241420cb0520ca  sensors/onion-preparation.zip
+477b42ac2c8044dc3ecec64a9f86429cc379f3e6ab7ba1a0c7ad24f48db45fa5  sensors/pepper-dicing.zip
+a6ad8727007b28c635c0626b0e250c34981e46ac681d2009b143ed5117b139bc  sensors/pepper-slicing.zip
+cc043779f7c28a786241a49ae79320384357a943629e721b34ba6e3a7e54bec3  source_preservation.json
+88a359a60c521ed841b89d99d807717527c0aa074c61f64e3e65ba85dd6e27a4  validate_release.py
+a2b23efc64a1b0320a3ed14bc10604c1b7ba09e5b7763ccf188ad260190e7555  validation.json
+7279f0d51cfc9ea5d01088aaa45a1777036da8300f0316462befad9a25bdbcab  videos/onion-preparation.mp4
+e164e7b06fa2f73e41f87eba48f59ef9b808a9b99516a9a02e6cda3afcf7558b  videos/pepper-dicing.mp4
+8a7efeadc9a2bd2cdac97c7f55df35ef6cb144b64c98d707251f5caae168f1da  videos/pepper-slicing.mp4

data/episodes.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cd1120d8395e43bd67a9bff6d570ef10323260f2e88f0e1641db67c8b9da99a
+size 54925242

dataset.json

@@ -0,0 +1,109 @@
+{
+  "schema_version": "1.0.0",
+  "name": "diffraction-egocentric-kitchen-sample",
+  "type": "human_egocentric_observation",
+  "release_status": "public_evaluation_sample",
+  "license": "cc-by-4.0",
+  "min_depth_confidence": 1,
+  "coordinate_frame": "ARKit camera-to-world, metres, xyzw quaternion; OpenCV unprojection requires diag(1,-1,-1) camera-axis conversion",
+  "pose_source": "device_vio_estimate",
+  "robot_actions": "absent",
+  "episodes": [
+    {
+      "capture_id": "onion-preparation",
+      "instruction": "Prepare and peel the onion on the cutting board using hands and a kitchen knife.",
+      "split": "sample",
+      "session_group": "kitchen_capture_group_unverified",
+      "video": "videos/onion-preparation.mp4",
+      "preview": "previews/onion-preparation.mp4",
+      "sensors": "sensors/onion-preparation.zip",
+      "signals": {
+        "quality": "annotations/onion-preparation/quality.parquet",
+        "tracks_raw": "annotations/onion-preparation/tracks_raw.parquet",
+        "tracks": "annotations/onion-preparation/tracks.parquet",
+        "arkit_poses": "annotations/onion-preparation/arkit_poses.parquet",
+        "frame_mapping": "annotations/onion-preparation/frame_mapping.parquet",
+        "camera_intrinsics": "annotations/onion-preparation/camera_intrinsics.parquet",
+        "device_imu": "annotations/onion-preparation/device_imu.parquet",
+        "hand_pose_raw": "annotations/onion-preparation/hand_pose_raw.parquet",
+        "hand_pose": "annotations/onion-preparation/hand_pose.parquet",
+        "hand_world_trajectory": "annotations/onion-preparation/hand_world_trajectory.parquet",
+        "contact": "annotations/onion-preparation/contact.parquet",
+        "contact_events": "annotations/onion-preparation/contact_events.parquet"
+      },
+      "frame_count": 4543,
+      "duration_s": 75.733333,
+      "width": 1920,
+      "height": 1440,
+      "nominal_fps": 59.9868,
+      "source_archive_sha256": "4a25f7c7080ee803ce6e9d4837ae2de68df296b37f60e37b43dce02ce04d87a4",
+      "video_sha256": "7279f0d51cfc9ea5d01088aaa45a1777036da8300f0316462befad9a25bdbcab",
+      "sensor_archive_sha256": "f3257b74551aed7ed90adffc7d2462a11234b984ad9d2b393c241420cb0520ca"
+    },
+    {
+      "capture_id": "pepper-slicing",
+      "instruction": "Prepare and slice the bell pepper on the cutting board using a kitchen knife.",
+      "split": "sample",
+      "session_group": "kitchen_capture_group_unverified",
+      "video": "videos/pepper-slicing.mp4",
+      "preview": "previews/pepper-slicing.mp4",
+      "sensors": "sensors/pepper-slicing.zip",
+      "signals": {
+        "quality": "annotations/pepper-slicing/quality.parquet",
+        "tracks_raw": "annotations/pepper-slicing/tracks_raw.parquet",
+        "tracks": "annotations/pepper-slicing/tracks.parquet",
+        "arkit_poses": "annotations/pepper-slicing/arkit_poses.parquet",
+        "frame_mapping": "annotations/pepper-slicing/frame_mapping.parquet",
+        "camera_intrinsics": "annotations/pepper-slicing/camera_intrinsics.parquet",
+        "device_imu": "annotations/pepper-slicing/device_imu.parquet",
+        "hand_pose_raw": "annotations/pepper-slicing/hand_pose_raw.parquet",
+        "hand_pose": "annotations/pepper-slicing/hand_pose.parquet",
+        "hand_world_trajectory": "annotations/pepper-slicing/hand_world_trajectory.parquet",
+        "contact": "annotations/pepper-slicing/contact.parquet",
+        "contact_events": "annotations/pepper-slicing/contact_events.parquet"
+      },
+      "frame_count": 3465,
+      "duration_s": 57.75,
+      "width": 1920,
+      "height": 1440,
+      "nominal_fps": 60.0,
+      "source_archive_sha256": "166703cbbc48dea22dbde6b7ef461c51199f244bb5d86a60fce1449066d062b4",
+      "video_sha256": "8a7efeadc9a2bd2cdac97c7f55df35ef6cb144b64c98d707251f5caae168f1da",
+      "sensor_archive_sha256": "a6ad8727007b28c635c0626b0e250c34981e46ac681d2009b143ed5117b139bc"
+    },
+    {
+      "capture_id": "pepper-dicing",
+      "instruction": "Dice the bell pepper strips on the cutting board using a kitchen knife.",
+      "split": "sample",
+      "session_group": "kitchen_capture_group_unverified",
+      "video": "videos/pepper-dicing.mp4",
+      "preview": "previews/pepper-dicing.mp4",
+      "sensors": "sensors/pepper-dicing.zip",
+      "signals": {
+        "quality": "annotations/pepper-dicing/quality.parquet",
+        "tracks_raw": "annotations/pepper-dicing/tracks_raw.parquet",
+        "tracks": "annotations/pepper-dicing/tracks.parquet",
+        "arkit_poses": "annotations/pepper-dicing/arkit_poses.parquet",
+        "frame_mapping": "annotations/pepper-dicing/frame_mapping.parquet",
+        "camera_intrinsics": "annotations/pepper-dicing/camera_intrinsics.parquet",
+        "device_imu": "annotations/pepper-dicing/device_imu.parquet",
+        "hand_pose_raw": "annotations/pepper-dicing/hand_pose_raw.parquet",
+        "hand_pose": "annotations/pepper-dicing/hand_pose.parquet",
+        "hand_world_trajectory": "annotations/pepper-dicing/hand_world_trajectory.parquet",
+        "contact": "annotations/pepper-dicing/contact.parquet",
+        "contact_events": "annotations/pepper-dicing/contact_events.parquet"
+      },
+      "frame_count": 2062,
+      "duration_s": 34.366667,
+      "width": 1920,
+      "height": 1440,
+      "nominal_fps": 60.0,
+      "source_archive_sha256": "fa7c7f295d02aab54901688f44d666283ba43f35a97a2b03c3ea7fe2351ce49b",
+      "video_sha256": "e164e7b06fa2f73e41f87eba48f59ef9b808a9b99516a9a02e6cda3afcf7558b",
+      "sensor_archive_sha256": "477b42ac2c8044dc3ecec64a9f86429cc379f3e6ab7ba1a0c7ad24f48db45fa5"
+    }
+  ],
+  "overview": "media/sample-overview.jpg",
+  "validation_report": "validation.json",
+  "huggingface_repo_id": "diffracting/egocentric-kitchen-sample"
+}

episodes/onion-preparation.json

@@ -0,0 +1,24 @@
+{
+  "schema_version": "1.0.0",
+  "capture_id": "onion-preparation",
+  "asset_id": "4a25f7c7080ee803",
+  "embodiment": "human_egocentric",
+  "dataset_type": "observational_demonstration",
+  "instruction": "Prepare and peel the onion on the cutting board using hands and a kitchen knife.",
+  "instruction_source": "assistant_video_inspection",
+  "session_group": "kitchen_capture_group_unverified",
+  "split": "sample",
+  "robot_actions": "absent",
+  "robot_state": "absent",
+  "annotation_accuracy": "not_independently_evaluated",
+  "privacy_review": "automated_full_video_face_scan; sampled_visual_inspection; no_human_attestation",
+  "annotation_review": "assistant_sampled_visual_review; independent_accuracy_not_measured",
+  "rights": {
+    "status": "owner_attested",
+    "training_use": true,
+    "public_sample_preparation": true,
+    "evidence": "Owner stated: the Stray Scanner recordings are owned and may be used for this training-data sample. The clay clip is not owned and is excluded. This is an owner attestation, not an independently verified signed legal release."
+  },
+  "pose_sample_fps": 15.0,
+  "license": "cc-by-4.0"
+}

episodes/pepper-dicing.json

@@ -0,0 +1,24 @@
+{
+  "schema_version": "1.0.0",
+  "capture_id": "pepper-dicing",
+  "asset_id": "fa7c7f295d02aab5",
+  "embodiment": "human_egocentric",
+  "dataset_type": "observational_demonstration",
+  "instruction": "Dice the bell pepper strips on the cutting board using a kitchen knife.",
+  "instruction_source": "assistant_video_inspection",
+  "session_group": "kitchen_capture_group_unverified",
+  "split": "sample",
+  "robot_actions": "absent",
+  "robot_state": "absent",
+  "annotation_accuracy": "not_independently_evaluated",
+  "privacy_review": "automated_full_video_face_scan; sampled_visual_inspection; no_human_attestation",
+  "annotation_review": "assistant_sampled_visual_review; independent_accuracy_not_measured",
+  "rights": {
+    "status": "owner_attested",
+    "training_use": true,
+    "public_sample_preparation": true,
+    "evidence": "Owner confirmed on 2026-09-08 that these Stray Scanner clips are owned and may be used; clay is excluded."
+  },
+  "pose_sample_fps": 15.0,
+  "license": "cc-by-4.0"
+}

episodes/pepper-slicing.json

@@ -0,0 +1,24 @@
+{
+  "schema_version": "1.0.0",
+  "capture_id": "pepper-slicing",
+  "asset_id": "166703cbbc48dea2",
+  "embodiment": "human_egocentric",
+  "dataset_type": "observational_demonstration",
+  "instruction": "Prepare and slice the bell pepper on the cutting board using a kitchen knife.",
+  "instruction_source": "assistant_video_inspection",
+  "session_group": "kitchen_capture_group_unverified",
+  "split": "sample",
+  "robot_actions": "absent",
+  "robot_state": "absent",
+  "annotation_accuracy": "not_independently_evaluated",
+  "privacy_review": "automated_full_video_face_scan; sampled_visual_inspection; no_human_attestation",
+  "annotation_review": "assistant_sampled_visual_review; independent_accuracy_not_measured",
+  "rights": {
+    "status": "owner_attested",
+    "training_use": true,
+    "public_sample_preparation": true,
+    "evidence": "Owner stated: the Stray Scanner recordings are owned and may be used for this training-data sample. The clay clip is not owned and is excluded. This is an owner attestation, not an independently verified signed legal release."
+  },
+  "pose_sample_fps": 15.0,
+  "license": "cc-by-4.0"
+}

observation_reader.py

@@ -0,0 +1,88 @@
+"""Standalone reader for Diffraction human egocentric RGB-D samples.
+
+Dependencies: numpy, polars, opencv-python. No pipeline or robot software needed.
+"""
+from __future__ import annotations
+import json
+import zipfile
+from pathlib import Path
+import cv2
+import numpy as np
+import polars as pl
+
+
+class ObservationDataset:
+    def __init__(self, root):
+        self.root = Path(root)
+        self.info = json.loads((self.root / "dataset.json").read_text(encoding="utf-8"))
+        self.episodes = {e["capture_id"]: e for e in self.info["episodes"]}
+
+    def table(self, capture_id, name):
+        e = self.episodes[capture_id]
+        return pl.read_parquet(self.root / e["signals"][name])
+
+    def frame(self, capture_id, frame_index):
+        """Return aligned native RGB, metric depth, confidence, K and camera pose.
+
+        ARKit pose is device VIO; no ground-truth or robot-action claim is implied.
+        Depth zero is invalid. Confidence filtering is explicit and reproducible.
+        """
+        e = self.episodes[capture_id]
+        if not isinstance(frame_index, int) or not 0 <= frame_index < e["frame_count"]:
+            raise IndexError(frame_index)
+        mapping = self.table(capture_id, "frame_mapping").filter(pl.col("rgb_frame_index") == frame_index).row(0, named=True)
+        source_frame_index = mapping["source_frame_index"]
+        intr = self.table(capture_id, "camera_intrinsics").filter(pl.col("source_frame_index") == source_frame_index).row(0, named=True)
+        pose = self.table(capture_id, "arkit_poses").filter(pl.col("t_s") == mapping["t_s"]).row(0, named=True)
+        cap = cv2.VideoCapture(str(self.root / e["video"]))
+        try:
+            cap.set(cv2.CAP_PROP_POS_FRAMES, frame_index)
+            ok, bgr = cap.read()
+        finally:
+            cap.release()
+        if not ok: raise RuntimeError(f"RGB decode failed at frame {frame_index}")
+        stem = f"{source_frame_index:06d}.png"
+        with zipfile.ZipFile(self.root / e["sensors"]) as z:
+            depth_raw = cv2.imdecode(np.frombuffer(z.read("depth/" + stem), np.uint8), cv2.IMREAD_UNCHANGED)
+            conf = cv2.imdecode(np.frombuffer(z.read("confidence/" + stem), np.uint8), cv2.IMREAD_UNCHANGED)
+        if depth_raw is None or conf is None or depth_raw.shape != conf.shape:
+            raise ValueError("Invalid depth/confidence pair")
+        depth = depth_raw.astype(np.float32) / 1000.
+        valid = (depth > 0) & (conf >= self.info["min_depth_confidence"])
+        depth[~valid] = 0
+        K = np.array([[intr["fx"], 0, intr["cx"]], [0, intr["fy"], intr["cy"]], [0, 0, 1]], np.float64)
+        return {"rgb": cv2.cvtColor(bgr, cv2.COLOR_BGR2RGB), "depth_m": depth,
+                "depth_valid": valid, "confidence": conf, "K_rgb": K,
+                "camera_pose_xyzw": np.array([pose[k] for k in ("tx", "ty", "tz", "qx", "qy", "qz", "qw")]),
+                "timestamp_s": mapping["t_s"], "sensor_timestamp_s": mapping["sensor_timestamp_s"],
+                "instruction": e["instruction"], "source_frame_index": source_frame_index, "rgb_frame_index": frame_index}
+
+    def annotations_near(self, capture_id, channel, timestamp_s, max_age_s=0.05):
+        """Sparse annotations stay absent beyond the caller's explicit age gate."""
+        df = self.table(capture_id, channel)
+        if df.is_empty(): return df
+        distance = (pl.col("t_s") - timestamp_s).abs()
+        nearby = df.filter(distance <= max_age_s)
+        if nearby.is_empty(): return nearby
+        nearest = nearby.select((pl.col("t_s") - timestamp_s).abs().arg_min()).item()
+        return nearby.filter(pl.col("t_s") == nearby["t_s"][nearest])
+
+
+def camera_points(frame):
+    """Valid depth pixels backprojected in OpenCV camera axes (+x right,+y down,+z forward)."""
+    depth, K = frame["depth_m"], frame["K_rgb"].copy()
+    dh, dw = depth.shape; rh, rw = frame["rgb"].shape[:2]
+    K[0, :] *= dw / rw; K[1, :] *= dh / rh
+    yy, xx = np.indices(depth.shape)
+    xyz = np.stack([(xx-K[0, 2])*depth/K[0, 0], (yy-K[1, 2])*depth/K[1, 1], depth], -1)
+    return xyz[frame["depth_valid"]]
+
+
+if __name__ == "__main__":
+    import argparse
+    p = argparse.ArgumentParser(); p.add_argument("root", type=Path)
+    args = p.parse_args(); ds = ObservationDataset(args.root)
+    for key, episode in ds.episodes.items():
+        frame = ds.frame(key, episode["frame_count"] // 2)
+        points = camera_points(frame)
+        print(key, frame["rgb"].shape, frame["depth_m"].shape, len(points), "valid camera points")

previews/metadata.jsonl

@@ -0,0 +1,3 @@
+{"file_name": "onion-preparation.mp4", "capture_id": "onion-preparation", "text": "Prepare and peel the onion on the cutting board using hands and a kitchen knife.", "duration_s": 75.733333, "native_rgb_frames": 4543, "split": "sample"}
+{"file_name": "pepper-slicing.mp4", "capture_id": "pepper-slicing", "text": "Prepare and slice the bell pepper on the cutting board using a kitchen knife.", "duration_s": 57.75, "native_rgb_frames": 3465, "split": "sample"}
+{"file_name": "pepper-dicing.mp4", "capture_id": "pepper-dicing", "text": "Dice the bell pepper strips on the cutting board using a kitchen knife.", "duration_s": 34.366667, "native_rgb_frames": 2062, "split": "sample"}

previews/onion-preparation.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37179bfaad964492247c48ef1418bdee941a62621d84796e4ca23144e5d195b2
+size 22188062