Add files using upload-large-folder tool

LICENSE

@@ -0,0 +1,9 @@
+MANUS-DexYCB is derived from DexYCB.
+
+DexYCB-derived source RGB images, crops, resized images, depth maps, and visual derivatives are governed by the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).
+
+Commercial use is not permitted unless separately authorized by the DexYCB rights holders.
+
+This repository does not grant any rights beyond those granted by the upstream DexYCB license. Users are responsible for complying with the upstream license, attribution requirements, change notices, and applicable laws.
+
+Please cite both MANUS / HanDrawer and the original DexYCB dataset when using this dataset.

README.md

@@ -0,0 +1,167 @@
+---
+pretty_name: "MANUS-DexYCB: DexYCB-derived Multimodal Annotated Naturalistic Hand Understanding Dataset"
+license: cc-by-nc-4.0
+tags:
+- image
+- depth-estimation
+- hand-pose
+- hand-gesture
+- 3d
+- mano
+- multimodal
+- multi-view
+- webdataset
+- computer-vision
+task_categories:
+- image-to-image
+- object-detection
+- image-classification
+configs:
+- config_name: dexycb
+  data_files:
+  - split: test
+    path: data/dexycb/test/*.tar
+gated: true
+extra_gated_prompt: >-
+  By requesting access, you acknowledge that MANUS-DexYCB is derived from
+  DexYCB and is licensed for non-commercial use under CC BY-NC 4.0.
+extra_gated_fields:
+  I agree to use this dataset for non-commercial purposes only: checkbox
+---
+
+# MANUS-DexYCB: DexYCB-derived Multimodal Annotated Naturalistic Hand Understanding Dataset
+
+**MANUS-DexYCB** is the DexYCB-derived subset of the Multimodal Annotated Naturalistic Hand Understanding (MANUS) dataset family. It is released as a source-specific repository because MANUS subsets are governed by different upstream licenses.
+
+This repository contains only the DexYCB-derived MANUS test split. HaGRID/HaGRIDv2-derived data is released separately as MANUS-HaGRID.
+
+## Dataset Features in This Release
+
+MANUS-DexYCB provides the following modalities where available:
+
+- full-scene RGB image;
+- full-scene depth map;
+- hand-region RGB crop;
+- hand-region depth map;
+- 2D bounding boxes for hand regions;
+- estimated MANO-style hand mesh annotations or related mesh metadata;
+- rendered mesh views from six canonical viewpoints: front, rear, left, right, top, and bottom;
+- primary and complementary mesh projection views where available;
+- companion text prompts describing image content, provided as the split-level prompt file listed below;
+- per-sample metadata recording source dataset, source id, split, available modalities, and license metadata.
+
+Text prompts are provided in `DexYCB_test_prompt.txt` as a companion split-level file.
+
+## Configs and Splits
+
+| Config/Split | Samples | WebDataset shards | Size |
+|---|---:|---:|---:|
+| `dexycb/test` | 7,836 | 7 | 5.58 GB |
+
+The canonical sample count is based on `DexYCB_test_OG_image_paths.txt`; other modalities are aligned to this OG image list.
+
+## Data Format
+
+Data is stored as uncompressed WebDataset tar shards. Files belonging to the same sample share the same prefix:
+
+```text
+dexycb_test_000001_processed_000001.json
+dexycb_test_000001_processed_000001.full_rgb.png
+dexycb_test_000001_processed_000001.full_depth.png
+dexycb_test_000001_processed_000001.hand_rgb.png
+dexycb_test_000001_processed_000001.hand_depth.png
+dexycb_test_000001_processed_000001.hand_bbox.json
+dexycb_test_000001_processed_000001.vertices.json
+dexycb_test_000001_processed_000001.mesh_front.png
+```
+
+The per-sample `.json` member records source and license metadata, source id, split, available modalities, and the number of files available for that sample. See `docs/data_structure.md` and `docs/annotation_format.md` for details.
+
+## Text Prompt Files
+
+The following companion prompt file has been added at the repository root:
+
+| File | Split | Records/Lines | Format | Notes |
+|---|---|---:|---|---|
+| `DexYCB_test_prompt.txt` | `dexycb/test` | 7,836 | TXT | One text prompt per line, aligned to the canonical DexYCB test OG image order. |
+
+Prompt files are optional companion annotations and are not packed into the WebDataset shards. Use this file when text conditioning or image-content descriptions are needed.
+
+## Metadata Files
+
+- `metadata/dexycb_manifest.parquet`
+- `metadata/dexycb_manifest.csv`
+- `metadata/dexycb_manifest.jsonl`
+- `metadata/source_licenses.csv`
+- `metadata/release_schema.json`
+- `metadata/release_report.json`
+- `metadata/checksums.sha256`
+- `DexYCB_test_prompt.txt`
+
+The public manifest files do not contain local absolute source paths. They contain portable source and audit metadata, including sample id, source id, source license, available modalities, and shard path.
+
+## Loading
+
+See `docs/loading_examples.md`.
+
+## Source Dataset Reference
+
+This release is derived from DexYCB. Please acknowledge the upstream DexYCB dataset when using MANUS-DexYCB.
+
+```bibtex
+@inproceedings{chao2021dexycb,
+  title={DexYCB: A benchmark for capturing hand grasping of objects},
+  author={Chao, Yu-Wei and Yang, Wei and Xiang, Yu and Molchanov, Pavlo and Handa, Ankur and Tremblay, Jonathan and Narang, Yashraj S and Van Wyk, Karl and Iqbal, Umar and Birchfield, Stan and others},
+  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
+  pages={9044--9053},
+  year={2021}
+}
+```
+
+## Citation
+
+If you use MANUS-DexYCB, please cite both the MANUS / HanDrawer work and the upstream DexYCB dataset.
+
+### MANUS / HanDrawer
+
+```bibtex
+@misc{fu2025handrawer,
+  title         = {HanDrawer: Leveraging Spatial Information to Render Realistic Hands Using a Conditional Diffusion Model in Single Stage},
+  author        = {Qifan Fu and Xu Chen and Muhammad Asad and Shanxin Yuan and Changjae Oh and Gregory Slabaugh},
+  year          = {2025},
+  eprint        = {2503.02127},
+  archivePrefix = {arXiv},
+  primaryClass  = {cs.CV},
+  doi           = {10.48550/arXiv.2503.02127},
+  url           = {https://arxiv.org/abs/2503.02127}
+}
+```
+
+### DexYCB
+
+Use the DexYCB citation listed in the source dataset reference above.
+
+## License
+
+MANUS-DexYCB is derived from DexYCB and is distributed under **Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)**. Commercial use is not permitted unless separately authorized by the DexYCB rights holders.
+
+This repository does not grant any rights beyond those granted by the upstream DexYCB license. Users are responsible for complying with the upstream license, attribution requirements, change notices, and applicable laws.
+
+## Intended Uses
+
+MANUS-DexYCB is intended for research on hand-object interaction, hand-aware image generation, 2D/3D hand understanding, depth estimation, and multimodal hand representation learning.
+
+## Out-of-Scope Uses
+
+This dataset should not be used for biometric identification, person recognition, surveillance, or privacy-invasive applications.
+
+## Privacy and Takedown
+
+MANUS-DexYCB is derived from images of people interacting with objects. Some images may include identifiable visual context. If you are a rights holder or data subject and request removal of a sample, please contact the dataset maintainers through the repository contact channel.
+
+## Related MANUS Repositories
+
+- MANUS landing page: `MANUS`
+- MANUS-HaGRID: HaGRID/HaGRIDv2-derived MANUS release
+
+This repository contains only DexYCB-derived data.

data/dexycb/test/dexycb-test-000000.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1908a73bafc58985dc0e039763edbc721aad6114a0ca76ec91aac5f24e783397
+size 974182400

data/dexycb/test/dexycb-test-000000.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efadd268be446b7dd6ebd2194f244fcb497a04ae575a1ced888ca21e2b271c8e
+size 260

data/dexycb/test/dexycb-test-000001.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c1c47433bdf2beebbd956bcf24c1fee9c6761eb9b0e3ea39fc83706f0d4dd40
+size 974673920

data/dexycb/test/dexycb-test-000001.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c266132d1e2151dbfda0381be5aba9bdc6dfc7ab5a6f885a0511a5fb5997125
+size 263

data/dexycb/test/dexycb-test-000002.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a13da605fa9158564057f8d5bffeaecc9b515164b3d25364c41ab97f3e0f2371
+size 973086720

data/dexycb/test/dexycb-test-000002.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57f70a879a0bf90ae423f2fc9d67ff4c382c0a0db11439ff45eaa6bdc9913375
+size 263

data/dexycb/test/dexycb-test-000003.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62835620e9df06369571312d595617c28df8ef39d179637cc256ce6e68e5a2be
+size 973342720

data/dexycb/test/dexycb-test-000003.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ce1ed9538c58d72e570965f23096fbd69e3074002ca243da7b1a1bce3ca9944
+size 263

data/dexycb/test/dexycb-test-000004.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3bb9cc490d46f4b90d407160f9305143d9979324604a49aed697ef717a61c21e
+size 975933440

data/dexycb/test/dexycb-test-000004.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2eba6b4fcde2cac935844ab134dbc6a0cb22d377b9c87ce20372db0c9e2613bf
+size 263

data/dexycb/test/dexycb-test-000005.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4fe2c6fc9a76b0c36f91b761791a3d65eb6ac213a9069e1f663ecb8795556b78
+size 972359680

data/dexycb/test/dexycb-test-000005.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:603874385ae8193ff938b3ba3140419fe699cdc9d27b1f8b33241cdc63f9434a
+size 263

data/dexycb/test/dexycb-test-000006.tar

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d47b3c8fd817d841b461db21e60d12972c17b91a1cc084e8ef920d35ecd0bb1
+size 149207040

data/dexycb/test/dexycb-test-000006.tar.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f2751af0f700862d010d5ba33b3e3ca4fb0911117667c9b39756889799228e91
+size 260

docs/annotation_format.md

@@ -0,0 +1,38 @@
+# MANUS-DexYCB Annotation Format
+
+Each WebDataset sample contains a `.json` metadata member plus the available multimodal files for that sample.
+
+## Per-Sample JSON Fields
+
+```json
+{
+  "manus_sample_id": "dexycb_test_000001_processed_000001",
+  "config": "dexycb",
+  "split": "test",
+  "source_dataset": "DexYCB",
+  "source_version": "DexYCB",
+  "source_split": "test",
+  "source_image_id": "000001_processed_000001",
+  "source_license_id": "cc-by-nc-4.0",
+  "source_license_name": "Creative Commons Attribution-NonCommercial 4.0 International",
+  "contains_source_pixels": true,
+  "processing_pipeline_version": "manus-dexycb-v1.0",
+  "available_files": ["full_rgb", "full_depth", "hand_rgb", "hand_depth", "vertices", "mesh_front"]
+}
+```
+
+## Modalities
+
+- `full_rgb.png`: original full-scene RGB image converted to PNG.
+- `full_depth.png`: full-scene depth map.
+- `hand_rgb.png`: hand-region RGB crop.
+- `hand_depth.png`: hand-region depth crop.
+- `hand_bbox.json`: hand-region bounding box for the RGB crop.
+- `hand_depth_bbox.json`: hand-region bounding box for the depth crop.
+- `vertices.json`: local hand mesh vertices and merged bounding-box metadata.
+- `mesh_front.png`, `mesh_back.png`, `mesh_left.png`, `mesh_right.png`, `mesh_top.png`, `mesh_bottom.png`: six canonical rendered mesh views.
+- `six_view_bbox_*.json`: bounding boxes for canonical rendered mesh views.
+- `extra_view_image_primary.png`, `extra_view_image_complementary.png`: selected primary/complementary projection views.
+- `extra_view_bbox_primary.json`, `extra_view_bbox_complementary.json`: bounding boxes for selected projection views.
+
+Not every future MANUS subset will expose the same modality set; check `available_files` and the manifest for each sample.

docs/data_structure.md

@@ -0,0 +1,38 @@
+# MANUS-DexYCB Release Structure
+
+This release is organized as one Hugging Face dataset repository for the DexYCB-derived MANUS subset.
+
+```text
+MANUS-DexYCB
+|-- README.md
+|-- LICENSE
+|-- data
+|   `-- dexycb
+|       `-- test/*.tar
+|-- metadata
+|   |-- dexycb_manifest.parquet
+|   |-- dexycb_manifest.csv
+|   |-- dexycb_manifest.jsonl
+|   |-- source_licenses.csv
+|   |-- release_schema.json
+|   |-- release_report.json
+|   `-- checksums.sha256
+|-- docs
+|   |-- data_structure.md
+|   |-- annotation_format.md
+|   |-- loading_examples.md
+|   `-- processing_pipeline.md
+`-- scripts
+    |-- audit_sources.py
+    `-- verify_checksums.py
+```
+
+## WebDataset Naming
+
+Tar shards are stored under `data/dexycb/test/` and named `dexycb-test-000000.tar`, `dexycb-test-000001.tar`, and so on.
+
+Within each tar, files belonging to the same sample share a common prefix such as `dexycb_test_000001_processed_000001`.
+
+## Public Metadata
+
+Public manifests do not contain local absolute source paths. They record portable sample metadata, available modalities, source dataset/license fields, and the relative shard path.

docs/loading_examples.md

@@ -0,0 +1,28 @@
+# Loading MANUS-DexYCB Examples
+
+## With WebDataset
+
+```python
+import json
+import webdataset as wds
+
+dataset = wds.WebDataset("data/dexycb/test/dexycb-test-{000000..000007}.tar")
+
+for sample in dataset:
+    meta_key = next(k for k in sample if k.endswith(".json"))
+    meta = json.loads(sample[meta_key].decode("utf-8"))
+    print(meta["manus_sample_id"], meta["available_files"])
+    break
+```
+
+## After Hugging Face Upload
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("QFun/MANUS-DexYCB", "dexycb", split="test", streaming=True)
+first = next(iter(ds))
+print(first.keys())
+```
+
+Official Hugging Face repository id: `QFun/MANUS-DexYCB`.

docs/processing_pipeline.md

@@ -0,0 +1,12 @@
+# MANUS-DexYCB Processing Pipeline
+
+The release is constructed from the local DexYCB source directory in read-only mode.
+
+1. Use `DexYCB_test_OG_image_paths.txt` as the canonical sample list.
+2. Map each OG image id to aligned depth, hand crop, hand-depth crop, bbox, mesh vertices, six-view mesh renderings, and projection-view files by deterministic filename rules.
+3. Filter all modalities to the canonical OG image ids.
+4. Package the aligned files into uncompressed WebDataset tar shards under `data/dexycb/test/`.
+5. Write public manifests without local absolute source paths.
+6. Write checksums for all release files.
+
+The source dataset directory is not modified by this pipeline.

metadata/checksums.sha256

@@ -0,0 +1,29 @@
+1908a73bafc58985dc0e039763edbc721aad6114a0ca76ec91aac5f24e783397  data/dexycb/test/dexycb-test-000000.tar
+efadd268be446b7dd6ebd2194f244fcb497a04ae575a1ced888ca21e2b271c8e  data/dexycb/test/dexycb-test-000000.tar.json
+8c1c47433bdf2beebbd956bcf24c1fee9c6761eb9b0e3ea39fc83706f0d4dd40  data/dexycb/test/dexycb-test-000001.tar
+7c266132d1e2151dbfda0381be5aba9bdc6dfc7ab5a6f885a0511a5fb5997125  data/dexycb/test/dexycb-test-000001.tar.json
+a13da605fa9158564057f8d5bffeaecc9b515164b3d25364c41ab97f3e0f2371  data/dexycb/test/dexycb-test-000002.tar
+57f70a879a0bf90ae423f2fc9d67ff4c382c0a0db11439ff45eaa6bdc9913375  data/dexycb/test/dexycb-test-000002.tar.json
+62835620e9df06369571312d595617c28df8ef39d179637cc256ce6e68e5a2be  data/dexycb/test/dexycb-test-000003.tar
+3ce1ed9538c58d72e570965f23096fbd69e3074002ca243da7b1a1bce3ca9944  data/dexycb/test/dexycb-test-000003.tar.json
+3bb9cc490d46f4b90d407160f9305143d9979324604a49aed697ef717a61c21e  data/dexycb/test/dexycb-test-000004.tar
+2eba6b4fcde2cac935844ab134dbc6a0cb22d377b9c87ce20372db0c9e2613bf  data/dexycb/test/dexycb-test-000004.tar.json
+4fe2c6fc9a76b0c36f91b761791a3d65eb6ac213a9069e1f663ecb8795556b78  data/dexycb/test/dexycb-test-000005.tar
+603874385ae8193ff938b3ba3140419fe699cdc9d27b1f8b33241cdc63f9434a  data/dexycb/test/dexycb-test-000005.tar.json
+4d47b3c8fd817d841b461db21e60d12972c17b91a1cc084e8ef920d35ecd0bb1  data/dexycb/test/dexycb-test-000006.tar
+f2751af0f700862d010d5ba33b3e3ca4fb0911117667c9b39756889799228e91  data/dexycb/test/dexycb-test-000006.tar.json
+138c1be13e186e3543f2f8fe3658ade02f53932c1dbe4ae1b5846c02f20b3095  DexYCB_test_prompt.txt
+f03ebe1ee9eb51c037b4f80f15ce65cf507979e13889688ca2569e4718e63912  docs/annotation_format.md
+a145a3510ace9346c4e70160822842df7dd8635ef91697dcd5c51e9403e0eb04  docs/data_structure.md
+9537e8b4d9c7735fa8f19d3e46083da43f9d40cd587687987f483f2a9ca2dd2b  docs/loading_examples.md
+39b9f59cd4958e374daa219639214880687c44fe3b06b634696f32d15bf75e00  docs/processing_pipeline.md
+f9e2bfaeceb9ef62f7208436c64d23e63dc7c23f1158882517dab90a1b5d85b2  LICENSE
+0470230ff83e1ce1dfb01413d24255217e0b3bd1b7db20877c0d2e63db85b647  metadata/dexycb_manifest.csv
+6fc58bad160ecb69e1fbb66526f8413944a172453cd7ae714c717d53e9c3afd4  metadata/dexycb_manifest.jsonl
+6f62e5d461d186e2e49b0c8c421807555fa608e25d989ff9bcff43f553b28940  metadata/dexycb_manifest.parquet
+5cefcb69164fdb43bde06ea1ca0f8460c61605589f5283c589c8283aeaded67e  metadata/release_report.json
+b4ca79ec8a42f4221303a2e08a2b9635d5531bf53e60c5cc6250e251335768a2  metadata/release_schema.json
+848bc824e993334c1dccf0c21f4c4c19c7b662e30c50181f60262398e1d0d9a6  metadata/source_licenses.csv
+e974ad80752e8e7826d6231c6318f59cb33adbddf7ddc7e43170584fd6471787  README.md
+1b9732b6bffc8a1384b9ce10d34942ea03a3b3622bece108a76f828ffd54f336  scripts/audit_sources.py
+77a53d968eccb04736fc62454a1ea8d55f412717c849020034ed7b03ab87b154  scripts/verify_checksums.py

metadata/dexycb_manifest.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f62e5d461d186e2e49b0c8c421807555fa608e25d989ff9bcff43f553b28940
+size 199719

metadata/release_report.json

@@ -0,0 +1,208 @@
+{
+  "created_by": "Codex data release preparation",
+  "dataset_name": "MANUS-DexYCB",
+  "dataset_family": "Multimodal Annotated Naturalistic Hand Understanding (MANUS)",
+  "sample_counts": {
+    "dexycb/test": 7836
+  },
+  "modality_presence": {
+    "dexycb/test": {
+      "extra_view_bbox_complementary": {
+        "present": 7836,
+        "missing": 0
+      },
+      "extra_view_bbox_primary": {
+        "present": 7836,
+        "missing": 0
+      },
+      "extra_view_image_complementary": {
+        "present": 7836,
+        "missing": 0
+      },
+      "extra_view_image_primary": {
+        "present": 7836,
+        "missing": 0
+      },
+      "full_depth": {
+        "present": 7836,
+        "missing": 0
+      },
+      "full_rgb": {
+        "present": 7836,
+        "missing": 0
+      },
+      "hand_bbox": {
+        "present": 7836,
+        "missing": 0
+      },
+      "hand_depth": {
+        "present": 7836,
+        "missing": 0
+      },
+      "hand_depth_bbox": {
+        "present": 7836,
+        "missing": 0
+      },
+      "hand_rgb": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_back": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_bottom": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_front": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_left": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_right": {
+        "present": 7836,
+        "missing": 0
+      },
+      "mesh_top": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_back": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_bottom": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_front": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_left": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_right": {
+        "present": 7836,
+        "missing": 0
+      },
+      "six_view_bbox_top": {
+        "present": 7836,
+        "missing": 0
+      },
+      "vertices": {
+        "present": 7836,
+        "missing": 0
+      }
+    }
+  },
+  "shards": [
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000000.tar",
+      "row_start": 0,
+      "row_end": 1271,
+      "sample_count": 1272,
+      "file_count": 30528,
+      "payload_bytes_estimate": 1073684905,
+      "size_bytes": 974182400
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000001.tar",
+      "row_start": 1272,
+      "row_end": 2536,
+      "sample_count": 1265,
+      "file_count": 30360,
+      "payload_bytes_estimate": 1073613286,
+      "size_bytes": 974673920
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000002.tar",
+      "row_start": 2537,
+      "row_end": 3816,
+      "sample_count": 1280,
+      "file_count": 30720,
+      "payload_bytes_estimate": 1073172646,
+      "size_bytes": 973086720
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000003.tar",
+      "row_start": 3817,
+      "row_end": 5098,
+      "sample_count": 1282,
+      "file_count": 30768,
+      "payload_bytes_estimate": 1073645190,
+      "size_bytes": 973342720
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000004.tar",
+      "row_start": 5099,
+      "row_end": 6348,
+      "sample_count": 1250,
+      "file_count": 30000,
+      "payload_bytes_estimate": 1073657851,
+      "size_bytes": 975933440
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000005.tar",
+      "row_start": 6349,
+      "row_end": 7636,
+      "sample_count": 1288,
+      "file_count": 30912,
+      "payload_bytes_estimate": 1073086132,
+      "size_bytes": 972359680
+    },
+    {
+      "config": "dexycb",
+      "split": "test",
+      "shard": "data/dexycb/test/dexycb-test-000006.tar",
+      "row_start": 7637,
+      "row_end": 7835,
+      "sample_count": 199,
+      "file_count": 4776,
+      "payload_bytes_estimate": 164770579,
+      "size_bytes": 149207040
+    }
+  ],
+  "total_tar_size_bytes": 5992785920,
+  "license": "Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)",
+  "source_image_note": "MANUS-DexYCB original-image sources are derived from DexYCB only.",
+  "citation": {
+    "title": "HanDrawer: Leveraging Spatial Information to Render Realistic Hands Using a Conditional Diffusion Model in Single Stage",
+    "arxiv_id": "2503.02127",
+    "doi": "10.48550/arXiv.2503.02127",
+    "url": "https://arxiv.org/abs/2503.02127"
+  },
+  "source_dataset_references": [
+    {
+      "name": "DexYCB",
+      "status_in_current_release": "included via the DexYCB-derived MANUS test split",
+      "bibtex_key": "chao2021dexycb",
+      "citation": {
+        "title": "DexYCB: A benchmark for capturing hand grasping of objects",
+        "authors": "Yu-Wei Chao, Wei Yang, Yu Xiang, Pavlo Molchanov, Ankur Handa, Jonathan Tremblay, Yashraj S. Narang, Karl Van Wyk, Umar Iqbal, Stan Birchfield, and others",
+        "venue": "Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition",
+        "pages": "9044--9053",
+        "year": 2021
+      }
+    }
+  ],
+  "public_manifest_contains_local_paths": false,
+  "source_preparation_note": "Source data were read from a local DexYCB preparation directory during release construction; the public manifest does not contain local absolute source paths."
+}

metadata/release_schema.json

@@ -0,0 +1,41 @@
+{
+  "format": "WebDataset source manifest v2",
+  "primary_key": "sample_id",
+  "dataset_name": "MANUS-DexYCB",
+  "dataset_family": "Multimodal Annotated Naturalistic Hand Understanding (MANUS)",
+  "configs": [
+    "dexycb"
+  ],
+  "splits": [
+    "test"
+  ],
+  "public_manifest_contains_local_paths": false,
+  "columns": [
+    "config",
+    "version",
+    "split",
+    "sample_id",
+    "manus_sample_id",
+    "source_dataset",
+    "source_version",
+    "source_split",
+    "source_id",
+    "source_image_id",
+    "numeric_id",
+    "gesture",
+    "source_namespace",
+    "source_license_id",
+    "source_license_name",
+    "contains_source_pixels",
+    "processing_pipeline_version",
+    "available_modalities",
+    "file_count",
+    "shard",
+    "notes"
+  ],
+  "notes": [
+    "This is the DexYCB-derived MANUS source-specific release prepared for Hugging Face.",
+    "Public manifests do not contain local absolute source paths.",
+    "The canonical sample count is based on DexYCB_test_OG_image_paths.txt."
+  ]
+}

metadata/source_licenses.csv

@@ -0,0 +1,2 @@
+source_dataset,source_version,license_id,license_name,license_url,contains_source_pixels,redistribution_basis
+DexYCB,DexYCB,cc-by-nc-4.0,Creative Commons Attribution-NonCommercial 4.0 International,https://creativecommons.org/licenses/by-nc/4.0/,true,upstream public non-commercial license

scripts/audit_sources.py

@@ -0,0 +1,18 @@
+from pathlib import Path
+import re
+import pandas as pd
+
+allowed_sources = {'DexYCB'}
+drive_path = re.compile(r"^[A-Za-z]:\\")
+for path in Path("metadata").glob('*manifest.parquet'):
+    df = pd.read_parquet(path)
+    if "source_dataset" not in df.columns:
+        raise ValueError(f"{path} is missing source_dataset")
+    bad = set(df["source_dataset"].dropna().unique()) - allowed_sources
+    if bad:
+        raise ValueError(f"{path} contains unexpected sources: {bad}")
+    for col in df.columns:
+        values = df[col].dropna().astype(str).head(1000)
+        if any(drive_path.match(v) or v.startswith("/data/") for v in values):
+            raise ValueError(f"{path} column {col} appears to contain local absolute paths")
+print('All MANUS-DexYCB manifests contain only DexYCB portable metadata.')

scripts/verify_checksums.py

@@ -0,0 +1,23 @@
+from pathlib import Path
+import hashlib
+
+root = Path(__file__).resolve().parents[1]
+checksum = root / "metadata" / "checksums.sha256"
+errors = []
+for line in checksum.read_text(encoding="utf-8").splitlines():
+    if not line.strip():
+        continue
+    expected, rel = line.split("  ", 1)
+    path = root / rel
+    h = hashlib.sha256()
+    with path.open("rb") as f:
+        for chunk in iter(lambda: f.read(8 * 1024 * 1024), b""):
+            h.update(chunk)
+    actual = h.hexdigest()
+    if actual != expected:
+        errors.append((rel, expected, actual))
+if errors:
+    for rel, expected, actual in errors:
+        print(f"MISMATCH {rel}: expected {expected}, got {actual}")
+    raise SystemExit(1)
+print("All checksums OK.")