I built a real-time ASL interpreter for the Gemma4 challenge, no cloud needed
Ngawang Tenzin developed a real-time American Sign Language (ASL) alphabet interpreter as part of the Gemma4 challenge, running entirely on local hardware without cloud dependency. The system uses MediaPipe for hand detection and cropping, then leverages the Gemma4:e4b model to identify ASL letters based on programmatic prompting without fine-tuning. Accuracy varies by letter, with some achieving 100% recognition while others like I, O, and P scored 0% in testing.
- ▪The ASL interpreter runs completely on-device using Gemma4:e4b, with no cloud API, subscription, or fine-tuning required.
- ▪MediaPipe preprocesses webcam input by detecting and cropping the hand to a 512×512 image, improving recognition accuracy by focusing the model solely on the hand.
- ▪The system uses detailed textual descriptions and disambiguation rules in the prompt to help Gemma4 recognize ASL letters it wasn't specifically trained on.
- ▪Evaluation was conducted using video frames and a batch testing pipeline, logging results and user feedback to measure per-letter accuracy.
- ▪Overall accuracy was 59.1%, with strong performance on letters like B, D, E, and Z, but consistent failures on I, O, and P.
Opening excerpt (first ~120 words) tap to expand
try { if(localStorage) { let currentUser = localStorage.getItem('current_user'); if (currentUser) { currentUser = JSON.parse(currentUser); if (currentUser.id === 3081552) { document.getElementById('article-show-container').classList.add('current-user-is-article-author'); } } } } catch (e) { console.error(e); } Ngawang Tenzin Posted on May 17 I built a real-time ASL interpreter for the Gemma4 challenge, no cloud needed #devchallenge #gemmachallenge #gemma #opensource Gemma 4 Challenge: Build With Gemma 4 Submission This is a submission for the Gemma 4 Challenge: Build with Gemma 4 What I Built A real-time American Sign Language (ASL) alphabet interpreter that runs 100% on your own machine - no api key, no cloud, no subscription.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at DEV.to (Top).
Discussion
0 commentsMore from DEV.to (Top)
