The timely response to traffic accidents is critical for reducing casualties and mitigating property damage. Traditional dashcams, while valuable for recording events, are inherently passive devices, requiring manual review and intervention to initiate any form of response. This paper presents DriveEYE, an intelligent system designed to augment conventional dashcams with advanced AI-powered capabilities, transforming them from simple recording devices into proactive safety tools. DriveEYE actively monitors the video feed in real-time to detect collision events, automatically analyzes the scene to gather crucial evidence, and instantly sends detailed emergency alerts. The system's prototype is developed and demonstrated using standard smartphone and laptop cameras to replicate the video input of a dashcam, showcasing its potential for integration into existing hardware. The proposed solution leverages a sophisticated pipeline: a continuous 30-second pre-crash buffer ensures critical lead-up footage is retained; sudden spikes in motion, detected via optical flow, trigger the incident protocol; a multi-threaded analysis then employs a specialized YOLOv8 model for license plate detection, the Google Gemini API for contextual scene summarization, and Pytesseract for Optical Character Recognition. All collected data—including an AI-generated summary, vehicle screenshots, license plate details, and simulated GPS coordinates—is compiled into a comprehensive PDF report and automatically dispatched to emergency contacts via SMS and email using the Twilio API and smtplib. This low-cost, integrated approach, built in Python, significantly reduces the time from incident occurrence to emergency response, offering a powerful and accessible method for enhancing driver safety and streamlining post-accident procedures.
Kale. et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: