Barbados Traffic Analysis Challenge
ML solution for predicting traffic congestion using video analysis. Implemented YOLO detection, VideoMAE embeddings, and ensemble models.
Computer Vision · VLM · Mobile App · Accessibility
Python, Computer Vision, VLMs, TTS, Flutter, FastAPI LinksAll projects
Overview
WrinnnAI is a real-time streaming mobile assistant designed for the visually impaired. It uses Computer Vision and Speech processing to provide environmental awareness and navigation assistance. Built between April and May 2025.
Key Features
- Environmental Description: Real-time scene understanding and narration using Vision-Language Models (VLMs)
- Obstacle Alerts: Automatic detection and warning for obstacles in the user’s path
- GPS Navigation: Voice-guided navigation with turn-by-turn directions
- Voice Commands: Hands-free interaction through natural language commands
- Low Latency Streaming: Optimized for real-time performance on mobile devices
Technologies
- Backend: Python, FastAPI, Computer Vision (OpenCV), Vision-Language Models
- Mobile App: Flutter for cross-platform deployment
- Speech: Text-to-Speech (TTS) for audio feedback
- AI Models: Transformers for image understanding and command parsing
Impact
This project aims to improve independence and safety for visually impaired individuals by providing real-time awareness of their surroundings through accessible audio feedback.