Add README
Browse files
README.md
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: other
|
| 3 |
+
task_categories:
|
| 4 |
+
- image-segmentation
|
| 5 |
+
- keypoint-detection
|
| 6 |
+
language:
|
| 7 |
+
- en
|
| 8 |
+
tags:
|
| 9 |
+
- synthetic
|
| 10 |
+
- internal
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
# test5555
|
| 14 |
+
|
| 15 |
+
Internal synthetic image dataset. **Not intended for redistribution or external use.**
|
| 16 |
+
|
| 17 |
+
## Format
|
| 18 |
+
|
| 19 |
+
- `data/bNNN.parquet` — one shard per source batch (~512 frames × 4 cameras)
|
| 20 |
+
- Columns: `id`, `split`, `batch`, `pose_index`, `camera`, `rgb`, `mask`, `keypoints`, `bones`
|
| 21 |
+
- `rgb` / `mask`: PNG bytes (not base64)
|
| 22 |
+
- `keypoints` / `bones`: JSON strings
|
| 23 |
+
- Resolution: 720×1280, black background
|
| 24 |
+
|
| 25 |
+
## Usage
|
| 26 |
+
|
| 27 |
+
```python
|
| 28 |
+
from datasets import load_dataset
|
| 29 |
+
ds = load_dataset("todo1111/test5555", data_files="data/*.parquet", split="train")
|
| 30 |
+
row = ds[0]
|
| 31 |
+
img = row["rgb"] # PIL via datasets Image(decode=True) if cast
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
+
## License
|
| 35 |
+
|
| 36 |
+
Private research use only. Do not redistribute.
|