Clarify dataset quality annotations and quantification; correct person ID

Clarify human review and metric-based failure segments, and report cleaned-release episode count, duration, object/person counts, and archive size.

Correct person ID 0014 to 014 in 2026-04-22_09-52-17_basketball_ball_pass_ankles_07 and update the release manifests. The other 54 files in that archive are unchanged; the release remains 4,135 episodes.

README.md

@@ -62,11 +62,29 @@ This dataset is ideal for training HOI reconstruction models like [CARI4D](https
 | Object mesh | `output_aligned.glb` metric textured object mesh in GLB format |
 | Object mesh symmetry | `output_symmetry.json` discrete/continuous rotational symmetries of the object mesh along principal axes |
 | Ground plane | `ground_plane.json` 4 coefficients of the estimated ground plane, Ax+By+Cz+D=0 |
-| Human QC results | `failure_segments.json` half-open ranges of frame indices where human/object pose annotation is deemed below standard |
+| Quality assessment results | `failure_segments.json` half-open ranges of frame indices flagged by human quality review or automated metric checks; see the explanation below |
+
+`failure_segments.json` records intervals flagged by human quality review or automated metric checks:
+
+- **Human QC:** Intervals where a reviewer visually inspected the rendered overlays and judged the pseudo-ground-truth annotations to be inaccurate.
+- **Chamfer distance:** Intervals flagged by discrepancies between reconstructed geometry and observed depth. These discrepancies may reflect annotation errors or depth noise; in practice, depth noise is frequently the underlying cause.
+- **Silhouette bounding box containment:** Intervals flagged by inconsistencies between projected geometry and segmentation-derived bounding boxes. These inconsistencies may reflect annotation errors or noisy segmentation masks.
+
+Metric-based flags indicate potential quality issues and should not be interpreted as confirmed annotation errors.
 
 ## Dataset Quantification <br>
-4,135 episodes, each containing the assets specified above. <br>
-The downloadable sequence archives total 4,503,845,089,280 bytes (approximately 4.504 TB), excluding release manifests.
+
+The statistics below cover only the cleaned release, excluding the held-out object sequences.
+
+| Property | Value |
+|----------|-------|
+| Episodes | 4,135 |
+| Total episode duration | ~29.144 hours |
+| Unique object IDs | 61 |
+| Unique person IDs | 22 |
+| Downloadable archive size | ~4.504 TB |
+
+Total duration counts each episode once, irrespective of the number of camera views. Archive size excludes release manifests and uses decimal TB.
 
 ## Reference(s): <br>
 Technical report is in progress. Code is integrated in the [video_to_data repo](https://github.com/nvidia-isaac/video_to_data/tree/main/reconstruction).

data/2026-04-22_09-52-17_basketball_ball_pass_ankles_07.tar

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e9c3cd0efd274ce6bf4850d39dcd973bf5243f50212d2d6791f538e7756157cb
+oid sha256:bc0b1a7048ad40ef76d08308bdbfa20b93087f5fc7af296030da666c531581cb
 size 1437501440

manifest/archives.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cf55ead7ed6cf82946d07a09fb005e267bf346daca0d45df0981007df062626e
-size 272114
+oid sha256:856e297f6c4f70a2c9b6041f327598c94c1bd935a1745cdb5056dc2973bd36ef
+size 272115

manifest/files.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a960da35b11afdb9549a43b4f5abb7664df590b6158150999f15a6ee950dfec
-size 7686397
+oid sha256:164c77b9f0b755ea58b9867987f83228a8fa6f852dbbcf36cdb8fac840ebfad4
+size 7685860

manifest/release.json

@@ -7,9 +7,9 @@
     "videos/*_stereo_camera_right.mp4",
     "commit.json"
   ],
-  "generated_at": "2026-09-18T00:04:49.596581+00:00",
+  "generated_at": "2026-09-18T05:46:25.677052+00:00",
   "layout": "one-uncompressed-tar-per-sequence",
-  "member_bytes": 4503583806980,
+  "member_bytes": 4503583806979,
   "member_count": 218519,
   "schema": "form-hoi-release-v1",
   "sequence_count": 4135