Logo

Real-Time BISINDO Detection

Real-Time BISINDO Detection is an interactive web application designed to detect and translate Indonesian Sign Language (BISINDO) in real time directly within the browser using machine learning and computer vision. Powered by Next.js, MediaPipe, and ONNX Runtime Web, it offers an accessible, privacy-focused, and seamless tool for sign language communication.

Tech Stack :

Description

Detailed information about this project.

BISINDO Recognition is a modern, client-side web application engineered to bridge the communication gap for the Deaf community by recognizing Indonesian Sign Language (BISINDO) in real time. Built with Next.js 16, React 19, and Tailwind CSS, the platform delivers a fast, responsive, and visually engaging user experience. By performing model inference locally using ONNX Runtime Web and Google MediaPipe Tasks Vision, all video processing occurs directly on the user's device, ensuring near-instant predictions, minimal latency, and full user data privacy.

Beyond single-gesture recognition, the application features an intelligent sequence detection engine that translates continuous hand movements into complete words and sentences. Designed with usability and accessibility at its core, the platform combines interactive live camera feeds with visual landmark overlays, real-time confidence indicators, and fluid animations powered by Framer Motion. This project represents a practical, accessible step forward in bringing AI-powered sign language translation to any modern web browser.

Features

Key features and capabilities.

In-Browser Inference

High-speed hand tracking and sign gesture classification running entirely in the browser via ONNX Runtime Web.

Smart Sequence Builder

Converts consecutive sign gestures into coherent words and sentences with automated buffer management.

Interactive Camera Overlay

Live video stream featuring real-time visual hand landmark overlays and performance monitoring.

Confidence Visualizer

Intuitive metrics display showing top predictions, confidence percentages, and recent history logs.

Fluid UI Experience

Sleek design with dark/light themes, kinetic typography, and responsive animations powered by Framer Motion.

TRY ADANN TRY ADANN