LinguaSignAI is an open-source project that translates American Sign Language (ASL) gestures into spoken words. It integrates two powerful techniques: keypoint-based recognition for speed and transfer learning for higher accuracy. By leveraging Google’s MediaPipe and pre-trained deep learning models, LinguaSignAI can adapt to diverse lighting conditions and handle complex gestures.
Check out the full source code and notebooks on GitHub. You’ll find data extraction scripts, model training pipelines, and instructions for setting up real-time detection. Whether you’re interested in expanding the sign vocabulary, fine-tuning the models, or optimizing performance on edge devices, your contributions are welcome!
Feel free to fork the repository, open issues, or submit pull requests. Together, we can make communication more inclusive and help bridge the gap between ASL users and the hearing world.
