AI-Powered Driving Learning Assistance
🏆 Winner – Theme: Road Safety at Hackathena 2024, Jyothi Engineering College, Cheruthuruthy
Drive.AI is an AI-powered driver learning assistance system designed to enhance road safety through real-time object detection, context-aware analysis, and instant voice feedback.
The system combines YOLOv5, OpenAI's language capabilities, and Google's Text-to-Speech to create an intelligent co-pilot that helps drivers make safer decisions while on the road.
Our project was developed during Hackathena 2024, where it was awarded Winner in the Road Safety category for its innovative approach to preventing accidents and improving driver awareness.
- Detects and analyzes pedestrians crossing in real time.
- Identifies nearby vehicles such as cars, motorcycles, and trucks.
- Recognizes animals crossing using models trained with Roboflow datasets.
- Works in diverse lighting conditions and urban/rural environments.
- Utilizes the YOLO (You Only Look Once) v5 model for fast and accurate object detection and classification.
- Generates a JSON output containing all detected objects with their confidence scores and classifications.
- Parses YOLO’s JSON output to extract relevant object information.
- Uses OpenAI's API to generate context-aware, instructional responses (e.g., "Slow down, pedestrian ahead on the left").
- Dynamic prompt creation for generating clear and actionable driving instructions.
- Integrates Google's Text-to-Speech (TTS) to instantly convert AI-generated responses into natural-sounding voice guidance.
- Provides real-time audio alerts directly to the driver, ensuring minimal distraction while driving.
- Camera Feed → Captures the driving environment in real time.
- YOLOv5 Detection → Processes frames to detect and classify objects.
- JSON Data Parsing → Extracts object details like type, position, and proximity.
- OpenAI API → Generates safety instructions based on detected objects.
- Google TTS → Converts instructions into voice guidance for immediate driver action.
- Event: Hackathena 2024
- Venue: Jyothi Engineering College, Cheruthuruthy
- Theme: Road Safety
- Award: Winner – Road Safety Category 🥇
- Recognized for innovative integration of AI and computer vision to proactively enhance road safety and prevent potential accidents.
- Object Detection: YOLOv5
- Data Annotation & Training: Roboflow
- Response Generation: OpenAI API
- Voice Output: Google Text-to-Speech API
- Backend: Python (Flask)
- Deployment: Local/Edge device processing for minimal latency
- GPS integration for location-aware safety instructions.
- Edge AI optimization for low-power embedded devices.
- Multilingual voice guidance for broader accessibility.
- Integration with AR HUDs for visual driving cues.