mysigner commited on
Commit
200c03d
·
verified ·
1 Parent(s): 7797deb

Delete README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -187
README.md DELETED
@@ -1,187 +0,0 @@
1
- ---
2
- license: cc-by-nc-sa-4.0
3
- language:
4
- - ms
5
- - zsm
6
- pretty_name: MySign — Malaysian Sign Language 3D Motion Capture
7
- size_categories:
8
- - 1K<n<10K
9
- task_categories:
10
- - other
11
- tags:
12
- - sign-language
13
- - bahasa-isyarat-malaysia
14
- - bim
15
- - malaysian-sign-language
16
- - motion-capture
17
- - 3d
18
- - fbx
19
- - skeletal-animation
20
- - gloss
21
- configs: []
22
- ---
23
-
24
- # MySign
25
-
26
- A 3D motion-capture dataset of Malaysian Sign Language (Bahasa Isyarat Malaysia, BIM), distributed as Filmbox (`.fbx`) skeletal animation files.
27
-
28
- > **Note**: this dataset is **not** loadable via `datasets.load_dataset(...)` as a tabular split, because the data itself is binary `.fbx`. `metadata.csv` is an index over the FBX files; consumers must read the file at `file_name` to access the motion data.
29
-
30
- ## Dataset Summary
31
-
32
- - **Modality**: 3D skeletal motion capture (`.fbx`)
33
- - **Language**: Malaysian Sign Language (BIM) — ISO `ms` / `zsm`
34
- - **Signers**: 5 (`Signer001` … `Signer005`)
35
- - **Sample**: one `.fbx` per `(signer, gloss, take)`
36
- - **License**: CC BY-NC-SA 4.0
37
- - **Source repository**: <https://huggingface.co/datasets/mysigner/MySign>
38
-
39
- ## Repository Structure
40
-
41
- ```
42
- MySign/
43
- ├── Signer001/
44
- │ ├── Above.fbx
45
- │ ├── Apologize.fbx
46
- │ ├── Apologize.001.fbx
47
- │ └── ...
48
- ├── Signer002/
49
- ├── Signer003/
50
- ├── Signer004/
51
- ├── Signer005/
52
- ├── metadata.csv # index over all .fbx files
53
- ├── croissant.json # Croissant 1.0 ML-dataset metadata (JSON-LD)
54
- ├── README.md
55
- └── .gitattributes
56
- ```
57
-
58
- ## `metadata.csv` Schema
59
-
60
- `metadata.csv` is a UTF-8 CSV with a header row. **Each row corresponds to exactly one `.fbx` file in the repository.**
61
-
62
- | Column | Type | Description |
63
- |-------------|---------|-------------|
64
- | `file_name` | string | Repository-relative path to the `.fbx`, e.g. `Signer001/Above.fbx`. |
65
- | `gloss` | string | Normalized gloss label (UPPERCASE). See [Gloss Normalization](#gloss-normalization). |
66
- | `signer_id` | string | One of `Signer001` … `Signer005`. |
67
- | `take` | integer | Take number for the `(signer_id, gloss)` pair. The original recording is `1`; Blender-style duplicate suffixes `.001`, `.002` map to takes `2`, `3`. |
68
-
69
- ### Example rows
70
-
71
- ```csv
72
- file_name,gloss,signer_id,take
73
- Signer001/Above.fbx,ABOVE,Signer001,1
74
- Signer001/Apologize.fbx,APOLOGIZE,Signer001,1
75
- Signer001/Apologize.001.fbx,APOLOGIZE,Signer001,2
76
- Signer002/Actor,_Actress.fbx,ACTOR / ACTRESS,Signer002,1
77
- Signer003/1-hr.fbx,1 HOUR,Signer003,1
78
- Signer004/Less (I).fbx,LESS(I),Signer004,1
79
- ```
80
-
81
- > **Path quoting**: some `file_name` values contain commas (e.g. `Signer002/Actor,_Actress.fbx`). The CSV is written with Python's default `csv.writer`, which quotes such fields automatically. Read it with any standard CSV parser (pandas, `csv.DictReader`).
82
-
83
- ### Gloss Normalization
84
-
85
- The `gloss` column is derived from each filename and normalized so that the same sign performed by different signers gets the same label. The normalization, applied by `generate_metadata_remote.py`, is:
86
-
87
- 1. Strip Blender suffix `.001` / `.002` (recorded into `take`).
88
- 2. Replace `_` and `-` with spaces.
89
- 3. Treat `,` and `;` as synonym separators, joined as `WORD_A / WORD_B`.
90
- 4. Uppercase.
91
- 5. Expand time-unit abbreviations: `HR→HOUR`, `MIN→MINUTE`, `MTH→MONTH`, `WK→WEEK`, `YR→YEAR`, `SEC→SECOND` (and plurals).
92
- 6. Normalize whitespace around parentheses: `WORD (X)` → `WORD(X)`.
93
- 7. Strip unbalanced trailing `)`.
94
- 8. Plural→singular merge: when the corpus contains both forms of a word (e.g. `COURSES` and `COURSE`), the plural is rewritten to the singular. A small block-list keeps semantically distinct plurals (`NEWS`, `SHORTS`, `MATHEMATICS`) as plurals.
95
- 9. A small hand-curated override map (currently: `COCHLEAR / IMPLANT` → `COCHLEAR IMPLANT`) fixes cases that no general rule can fix safely.
96
-
97
- A full audit of the plural→singular merges is printed by the script every time it runs; check that list before publishing.
98
-
99
- ## `.fbx` Files
100
-
101
- - Format: Autodesk Filmbox (binary `.fbx`).
102
- - One file per take.
103
- - The `file_name` column of `metadata.csv` is the canonical pointer to each file. The same path can be resolved as `https://huggingface.co/datasets/mysigner/MySign/resolve/main/<file_name>`.
104
-
105
- ## Croissant Metadata
106
-
107
- A Croissant 1.0 (JSON-LD) description is provided as `croissant.json`. It declares:
108
-
109
- - The repository as a `cr:FileObject` (`encodingFormat: git+https`).
110
- - `metadata.csv` as a `cr:FileObject` of type `text/csv`.
111
- - All `Signer*/*.fbx` files as a `cr:FileSet` named `fbx-files` (`encodingFormat: model/vnd.fbx`).
112
- - A `cr:RecordSet` named `signs` exposing the four columns above. The `file_name` field carries `references: fbx-files`, which tells Croissant consumers the value is a path into the FBX FileSet.
113
-
114
- **Limitation**: Croissant 1.0 has no native semantics for FBX. `mlcroissant` will treat every FBX as an opaque binary resource — it will not parse skeletons, animation curves, or any FBX-internal structure. Consumers must use a real FBX library (e.g. Autodesk FBX SDK, Blender, [`pyfbx`](https://pypi.org/project/pyfbx/)) to read the motion data.
115
-
116
- ## Usage
117
-
118
- ### List records via `metadata.csv`
119
-
120
- ```python
121
- import pandas as pd
122
-
123
- df = pd.read_csv(
124
- "https://huggingface.co/datasets/mysigner/MySign/resolve/main/metadata.csv"
125
- )
126
- print(df.head())
127
- print(df["signer_id"].value_counts().sort_index())
128
- print(df["gloss"].nunique(), "unique glosses")
129
- ```
130
-
131
- ### Download a specific FBX
132
-
133
- ```python
134
- from huggingface_hub import hf_hub_download
135
-
136
- local_path = hf_hub_download(
137
- repo_id="mysigner/MySign",
138
- filename="Signer001/Above.fbx",
139
- repo_type="dataset",
140
- )
141
- print(local_path) # local cache path; open with your FBX library
142
- ```
143
-
144
- ### Inspect via Croissant (mlcroissant)
145
-
146
- ```python
147
- import mlcroissant as mlc
148
-
149
- ds = mlc.Dataset(
150
- "https://huggingface.co/datasets/mysigner/MySign/resolve/main/croissant.json"
151
- )
152
- print(ds.metadata.name, "—", ds.metadata.description[:80])
153
-
154
- for i, rec in enumerate(ds.records(record_set="signs")):
155
- print(rec)
156
- if i >= 4:
157
- break
158
- ```
159
-
160
- `ds.records("signs")` yields one Python dict per `.fbx`, with `file_name`, `gloss`, `signer_id`, `take`. The FBX bytes are not auto-loaded — read them yourself from `file_name`.
161
-
162
- ## License
163
-
164
- This dataset is released under [Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/). You may share and adapt the work for non-commercial purposes, provided you give appropriate credit and distribute derivative works under the same license.
165
-
166
- ## Citation
167
-
168
- ```bibtex
169
- @misc{mysign2026,
170
- title = {MySign: A High-Fidelity Motion-Capture Dataset for 3D Sign Generation in Bahasa Isyarat Malaysia},
171
- author = {{mysigner}},
172
- year = {2026},
173
- howpublished = {Hugging Face Datasets},
174
- url = {https://huggingface.co/datasets/mysigner/MySign},
175
- note = {CC BY-NC-SA 4.0}
176
- }
177
- ```
178
-
179
- (Replace with the canonical citation when a paper or technical report becomes available.)
180
-
181
- ## Limitations
182
-
183
- - **Not auto-loadable**: Hugging Face's Dataset Viewer cannot render `.fbx`. The Viewer will not work for this dataset, and `datasets.load_dataset("mysigner/MySign")` will not produce a usable split. Use the workflows above instead.
184
- - **Gloss is filename-derived**: glosses come from filenames written by signers/annotators with slightly different conventions, then normalized. Some collisions may still exist; run the script's plural-merge audit before relying on a particular gloss inventory.
185
- - **Five signers**: signer-conditioned models trained on MySign will have limited speaker coverage.
186
- - **Take semantics are best-effort**: the `take` column distinguishes Blender-duplicated files (`.001`, `.002`) from the original recording, but it does not encode whether a take was a clean recording or a retry.
187
- - **No segmentation, no transcription**: each FBX is one isolated sign. The dataset does not include continuous-signing video, glossed sentences, or non-manual annotations.