Thai-Sign-Language 💭

This project aims to develop a Thai Sign Language detection using a Transformer-based model.

code </>

Model Files

• model.tflite: This model was trained using the TSL-transformer-training.ipynb notebook.
• model-withflip.tflite: This model was trained with additional data augmentation (flipping) using the TSL-withflip-transformer-training.ipynb notebook.

Data Processing

• parquetdata_visualize.ipynb: This notebook explains how the data was processed and visualized before being used for model training.

Detectable signs

• The list of detectable signs can be found in the actions.txt.
• For examples of Thai signs and their corresponding actions, you can visit https://www.th-sl.com/?openExternalBrowser=1.

app 🚀

You can try the thai sign language detection app through this link:

Link🔗 : Thai Sign Language Detection App

Alternatively, you can clone the repository and run the app locally to try real-time thai sign language detection app:

streamlit run real_time_app.py

dataset 🗃️

• numpy file : https://github.com/Annerez/Mediapipe_ThaiHandSign?fbclid=IwAR1OPuXqOO-M6qd8OI24fogd3GSZ_E1hA42iUwLtAGltuCNthxz7OmO1Ltg
• parquet file : https://drive.google.com/drive/folders/1f2uqMtYWMfLyeFzRrC0lwu0jG7ItVYsu?usp=sharing
• parquet file with flip : https://drive.google.com/drive/folders/1SGlHoKKlsNQA-vk4_sXyDML6Lo-PX5YR?usp=sharing

more info

For a detailed walkthrough of the project, including how to improve it, you can read my article on Medium: https://medium.com/@feat.meprxxw/thai-sign-language-detection-07a3d864a3b4