🚌 Automatic Bus Announcement System | Transit Vision 🚏

An AI-powered real-time bus arrival announcement system built using Python, Flask, OpenCV, YOLOv8 Object Detection, OCR (Google Vision API), and SORT Tracking Algorithm.

---

📌 Project Overview

Transit Vision is an Automatic Bus Announcement System designed for bus stands to assist passengers by announcing incoming buses. The system uses a live camera feed to detect KSRTC buses, recognize their route boards, and announce bus arrival information with voice output.

This is especially useful for visually impaired passengers and for crowded or large bus stands.

---

🚀 Features

• ✅ Real-time Bus Detection using YOLOv8
• ✅ Route Board Text Extraction using Google Vision OCR API
• ✅ Unique Bus Tracking using SORT (Simple Online Realtime Tracking)
• ✅ Text-to-Speech Bus Announcements
• ✅ Bus Arrival History Log with timestamp
• ✅ Live Web Dashboard built with Flask + HTML/CSS/JavaScript
• ✅ Start/Stop detection from the frontend
• ✅ Scrollable detection history (Bus Log)
• ✅ Graceful shutdown of processes and resources

---

🖥️ Technologies Used

• Python
• Flask (Web Framework)
• OpenCV (Video Streaming & Image Processing)
• YOLOv8 (Bus Detection)
• Google Vision API (OCR for Bus Route Text)
• SORT Algorithm (Object Tracking)
• HTML / CSS / JavaScript (Frontend)
• Threading and Queues (for parallel frame processing and audio playback)

---

⚙️ How It Works

1. Live camera feed captures frames.
2. YOLOv8 model detects buses and name boards.
3. OCR extracts Malayalam text from detected name boards.
4. SORT assigns a unique ID to each bus for tracking across frames.
5. For each unique bus detected with a new route text, the system:
   • Adds it to arrival history
   • Announces the bus arrival using TTS (Text-to-Speech)
6. Frontend shows:
   • Live video feed
   • Current bus being detected
   • Real-time bus arrival history (log)
   • Detection start/stop control buttons

---

📸 Screenshots

---

📂 Project Structure (Key Files/Folders)

app.py # Main Flask app functions.py # Detection logic, OCR, tracking, history, TTS templates/ # HTML frontend pages (index.html, golive.html, etc) static/ # CSS, JS, images history.json # Persistent Bus Arrival History

---

🧠 Algorithms Used

• YOLOv8 → Bus and Board Detection
• Google Vision API OCR → Text extraction from bus boards
• SORT (Kalman Filter + Hungarian Algorithm) → Object Tracking across frames
• TTS (Text-to-Speech) → Bus arrival audio announcements

---

🙋🏻‍♂️ Team Contribution / Customization Done

• ✅ Designed full system flow from live camera to announcement.
• ✅ Customized YOLOv8 model for KSRTC bus detection.
• ✅ Integrated Google Vision OCR for Malayalam text reading.
• ✅ Implemented SORT tracking for unique bus ID assignment.
• ✅ Developed real-time frontend dashboard with history log.
• ✅ Managed server-side threading, API rate control, and graceful shutdown.
• ✅ Implemented history.json file saving for persistent log.

---

👨‍💻 Authors

• Alfred Antony
• Anabel George
• Fabiya Philomina M J

---