| ---
|
| license: cc-by-4.0
|
| tags:
|
| - active
|
| - curriculum
|
| - legal
|
| - lmdb
|
| - mixup-cutmix
|
| - moderate
|
| - semi-auto
|
| - sensor-fusion
|
| - stratified-90-10
|
| ---
|
|
|
| # dataset.py
|
|
|
| ## Dataset Summary
|
|
|
| A **legal** dataset with **sensor fusion** modality, stored in **lmdb** format.
|
|
|
| ## Preprocessing & Augmentation
|
|
|
| - **Preprocessing**: curriculum
|
| - **Augmentation**: mixup cutmix
|
|
|
| ## Splits & Sampling
|
|
|
| - **Split strategy**: stratified 90 10
|
| - **Sampling**: active
|
|
|
| ## Quality & Labeling
|
|
|
| - **Quality filtering**: moderate
|
| - **Labeling**: semi auto
|
|
|
| ## Files
|
|
|
| - `dataset.py` — main artifact of this repository
|
|
|
| ## License
|
|
|
| See the license field above.
|
|
|
