| ---
|
| license: cc-by-4.0
|
| tags:
|
| - adaptive
|
| - art
|
| - auto-ml
|
| - balanced
|
| - hdf5
|
| - heavy
|
| - manual
|
| - stratified-90-10
|
| - video-text
|
| ---
|
|
|
| # build_dataset.py
|
|
|
| ## Dataset Summary
|
|
|
| A **art** dataset with **video text** modality, stored in **hdf5** format.
|
|
|
| ## Preprocessing & Augmentation
|
|
|
| - **Preprocessing**: auto ml
|
| - **Augmentation**: heavy
|
|
|
| ## Splits & Sampling
|
|
|
| - **Split strategy**: stratified 90 10
|
| - **Sampling**: balanced
|
|
|
| ## Quality & Labeling
|
|
|
| - **Quality filtering**: adaptive
|
| - **Labeling**: manual
|
|
|
| ## Files
|
|
|
| - `build_dataset.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
