Ankara University Turkish Sign Language Dataset
The Ankara University Turkish Sign Language Dataset (AUTSL) is a large-scale, multimode dataset that contains isolated Turkish sign videos. It contains 226 signs that are performed by 43 different signers. There are 38,336 video samples in total. The samples are recorded using Microsoft Kinect v2 in RGB, depth and skeleton formats. The videos are provided at a resolution of 512×512. The skeleton data contains spatial coordinates, i.e. (x, y), of the 25 junction points on the signer body that are aligned with 512×512 data.
Source: AUTSL Dataset
Variants: AUTSL
This dataset is used in 1 benchmark:
| Task | Model | Paper | Date |
|---|---|---|---|
| Sign Language Recognition | HWGAT | Hierarchical Windowed Graph Attention Network … | 2024-07-19 |
| Sign Language Recognition | MViT-SLR | Fine-tuning of sign language recognition … | 2023-02-15 |
| Sign Language Recognition | SAM-SLR (RGB-D) | Skeleton Aware Multi-modal Sign Language … | 2021-03-16 |
| Sign Language Recognition | CNN+FPM+BLSTM+Attention (RGB-D) | AUTSL: A Large Scale Multi-modal … | 2020-08-03 |
Recent papers with results on this dataset: