Sign in Subscribe
Artificial intelligence

Breakthrough in Sign Language Recognition Using Artificial Intelligence

Strategic Insights

— 1 min read

Researchers from the Nelson Mandela African Institution of Science and Technology have made significant progress in using artificial intelligence to recognize sign language. By leveraging computer vision and machine learning algorithms, they have developed a model that can accurately identify sign language with an accuracy rate of 94%. The study aimed to bridge the communication gap between speech-impaired populations and those without impairment, where most people are unaware of the sign language used by speech-impaired individuals. Using mobile phone selfie cameras, the researchers collected Tanzania Sign Language datasets to investigate the performance of deep learning algorithms that capture spatial and temporal relationships features of video frames.

Key Takeaways:

  • Researchers from the Nelson Mandela African Institution of Science and Technology have developed a model that can accurately identify sign language with an accuracy rate of 94%.
  • The study used Tanzania Sign Language datasets collected using mobile phone selfie cameras to investigate the performance of deep learning algorithms that capture spatial and temporal relationships features of video frames.
  • The proposed CNN-GRU model with ELU activation function is proposed to enhance learning efficiency and performance, achieving a 94% accuracy rate compared to 93% for the standard CNN-GRU model and CNN-LSTM.
  • The study evaluated performance of the proposed model in a signer-independent setting, where the results varied significantly across individual signers, with the highest accuracy reaching 66%.
  • The research concluded that more effort is required to improve signer independence performance, including the challenges of hand dominance by optimizing spatial features.
  • The study highlights the potential of artificial intelligence in bridging the communication gap between speech-impaired populations and those without impairment.

Statistics:

  • 94% accuracy rate achieved by the proposed CNN-GRU model with ELU activation function.
  • 93% accuracy rate achieved by the standard CNN-GRU model and CNN-LSTM.
  • 66% highest accuracy rate achieved in a signer-independent setting.
  • 472 Tanzania Sign Language datasets were collected using mobile phone selfie cameras.
  • The study used CNN-LSTM and CNN-GRU architectures to investigate the performance of deep learning algorithms.

Sources:

  • "Efficient spatio-temporal modeling for sign language recognition using CNN and RNN architectures." Frontiers in Artificial Intelligence, 2025,8.
  • Kasian Myagila, School of Computation and Communication Science and Engineering, Nelson Mandela African Institution of Science and Technology, Arusha, Tanzania.
  • Devotha Godfrey Nyambo and Mussa Ally Dida, additional authors on the research.

Read more