BISINDO Detection
BISINDO Detection is a browser-based application that detects and translates Indonesian Sign Language in real time using computer vision. Powered by Next.js, MediaPipe, and ONNX Runtime Web, it provides privacy-focused, low-latency sign language recognition.
Description
Detailed information about this project.
BISINDO Detection is a client-side web application built with Next.js, MediaPipe, and ONNX Runtime Web to translate Indonesian Sign Language (BISINDO) in real time. All video processing and model inference occur locally inside the user's browser, ensuring instant gesture prediction while preserving complete data privacy. The platform features hand landmark tracking, real-time confidence indicators, and a sequence detection engine that translates continuous gestures into words and sentences.
Features
Key features and capabilities.
Browser Inference
Run sign detection locally using ONNX Runtime Web.
Hand Landmark
Realtime visual tracking of key hand joint points.
Sequence Engine
Translate consecutive sign gestures into full sentences automatically.
Confidence Metrics
Display real-time gesture prediction probability and confidence scores.
Documentation
A brief overview of each page in this project.
Home Page
Main landing page with interactive floating sign cards.
BISINDO Detection
Real-time webcam sign detection interface with pose landmarks.
About System
System architecture, YOLOv11m-Pose model specs, and performance metrics.