Project Description
I’m developing a tool that recognises Indian Sign Language (ISL) from live camera input and instantly translates the signs on-screen. The goal is to make classroom and self-study environments more inclusive, so accuracy, speed, and an intuitive user experience matter more to me than flashy extras.
Here’s what I need from you:
• A working real-time detection and translation engine trained specifically on ISL datasets.
• Clean, well-commented code (Python + OpenCV / MediaPipe / TensorFlow or comparable libraries are fine) that I can retrain with new samples later.
• A lightweight interface—desktop or web—where a learner can see their signing mirrored, along with immediate text output.
• A short guide outlining model architecture, dataset sources, and steps to deploy on Windows or Linux.
Acceptance will be based on:
1. Live video accuracy that comfortably exceeds 90 % on a test set we’ll agree on.
2. Latency low enough to keep conversation flow natural (target <300 ms per frame).
3. Source files delivered in a repo I can run with a single setup script.
If this lines up with your expertise in computer vision and sign language tech, I’m ready to review your approach and timeline.