A New Roadway Eventual Obstacle Detection System Based on Computer Vision

Abstract

A new roadway eventual obstacle detection system based on computer vision is described and evaluated. This system uses low-cost hardware and open-source software to detect and classify moving elements in roads using infra-red and colour video images as input data. This solution represents an important advancement to prevent road accidents due to eventual obstacles which have considerably increased in the past decades, mainly with wildlife. The experimental evaluation of the system demonstrated that the proposed solution detects and classifies correctly different types of moving obstacles on roads, working robustly under different weather and illumination conditions.

Deep learning–based eye tracking system to detect distracted driving

“…However, drivers often cannot focus on multiple pieces of information simultaneously, thereby increasing safety risks [21]. Target detection algorithms have emerged as a mainstream method for sensing the driving environment [22][23][24][25][26]. In 2010, Mnih and Hinton [27] applied deep learning techniques to two datasets of remote sensing images, extracting road features for training with promising results.…”