Update README for full-data release layout
Browse files
README.md
CHANGED
|
@@ -28,12 +28,12 @@ scripts/
|
|
| 28 |
extract_vlac2_release_frames.py
|
| 29 |
|
| 30 |
data/
|
| 31 |
-
vlac2_release_data_part01_arx_group_a.tar.zst
|
| 32 |
-
vlac2_release_data_part02_arx_group_b.tar.zst
|
| 33 |
-
vlac2_release_data_part03_arx_group_c.tar.zst
|
| 34 |
-
vlac2_release_data_part04_droid_group_a.tar.zst
|
| 35 |
-
vlac2_release_data_part05_droid_group_b.tar.zst
|
| 36 |
-
vlac2_release_data_part06_droid_group_c.tar.zst
|
| 37 |
vlac2_release_data_part07_other_sources.tar.zst
|
| 38 |
```
|
| 39 |
|
|
@@ -45,7 +45,7 @@ data/
|
|
| 45 |
bash scripts/unpack_data.sh /path/to/data
|
| 46 |
```
|
| 47 |
|
| 48 |
-
|
| 49 |
|
| 50 |
### 2. Extract frames for a selected benchmark JSON
|
| 51 |
|
|
@@ -73,4 +73,4 @@ Replace `__VLAC2_FRAMES_ROOT__` with the absolute path to the extracted-frame di
|
|
| 73 |
* This release contains the complete raw-data archive set used by the VLAC-Cut benchmark release workflow.
|
| 74 |
* The `benchmark_style_all/all` split is larger than the benchmark evaluation subsets and is rebuilt from the full annotation PKL source tree.
|
| 75 |
* The benchmark JSON files are pre-generated and do not need to be rebuilt for standard use.
|
| 76 |
-
*
|
|
|
|
| 28 |
extract_vlac2_release_frames.py
|
| 29 |
|
| 30 |
data/
|
| 31 |
+
vlac2_release_data_part01_arx_group_a.tar.zst.part-*
|
| 32 |
+
vlac2_release_data_part02_arx_group_b.tar.zst.part-*
|
| 33 |
+
vlac2_release_data_part03_arx_group_c.tar.zst.part-*
|
| 34 |
+
vlac2_release_data_part04_droid_group_a.tar.zst.part-*
|
| 35 |
+
vlac2_release_data_part05_droid_group_b.tar.zst.part-*
|
| 36 |
+
vlac2_release_data_part06_droid_group_c.tar.zst.part-*
|
| 37 |
vlac2_release_data_part07_other_sources.tar.zst
|
| 38 |
```
|
| 39 |
|
|
|
|
| 45 |
bash scripts/unpack_data.sh /path/to/data
|
| 46 |
```
|
| 47 |
|
| 48 |
+
Several large archives are distributed as multi-part files to satisfy Hub file-size limits. `unpack_data.sh` detects these parts automatically and streams them into a single extraction pass.
|
| 49 |
|
| 50 |
### 2. Extract frames for a selected benchmark JSON
|
| 51 |
|
|
|
|
| 73 |
* This release contains the complete raw-data archive set used by the VLAC-Cut benchmark release workflow.
|
| 74 |
* The `benchmark_style_all/all` split is larger than the benchmark evaluation subsets and is rebuilt from the full annotation PKL source tree.
|
| 75 |
* The benchmark JSON files are pre-generated and do not need to be rebuilt for standard use.
|
| 76 |
+
* The large raw-data archives are distributed as split `.part-*` files in `data/`; these are unpacked transparently by the provided script.
|