the future of sign language interpretation

January 4, 2024

Limitations of SENYAS

While SENYAS successfully addressed the need for a Filipino Sign Language (FSL) recognition system, several limitations were encountered during development:

  1. Limited Dataset:

    • The lack of a publicly accessible FSL dataset meant the researchers had to create their own dataset. This process was time-consuming, and the dataset was relatively small, consisting of only 10 dynamic and 5 static gestures.
    • Regional variations of FSL, such as those from Southern Luzon or Eastern Visayas, were not represented, meaning the model may not generalize well to these dialects.
  2. Restricted Sign Categories:

    • The system is limited to detecting individual signs, rather than full sentences or phrases, which limits its utility in real-world conversations.
    • Only hand and pose landmarks were used for gesture recognition, and facial landmarks were omitted to keep the model lightweight, but this reduces accuracy for signs that involve facial expressions.
  3. Web Application Constraints:

    • The application is currently hosted on a cloud server in Japan, which could introduce latency issues for users in the Philippines, especially in areas with unstable internet connectivity.

Moving Forward

To enhance the SENYAS system, the following steps could be taken:

  1. Expanding the Dataset: A larger, more diverse dataset including regional dialects and additional gestures would make the model more robust and versatile.
  2. Improving Gesture Range: By expanding the system to recognize complete phrases and sentences, SENYAS could be used in a wider range of scenarios, such as interpreting conversations in real time.
  3. Face and Gesture Integration: Future iterations of the model could include facial landmarks to improve accuracy for signs that rely heavily on facial expressions.
  4. Optimizing the Web Application: Hosting the application on local servers would minimize latency and improve the user experience for those in the Philippines.

By addressing these limitations, SENYAS can become an even more powerful tool for FSL recognition and expand its impact within the deaf community.