Project Abstract

The advancement of autonomous driving and new ADAS safety systems has propelled the demand for robust and efficient real-time object detection systems within automotive platforms. However, object-detection algorithms struggle with detection in low light environments. Systems such as Adaptive Driving Beam (ADB), improve driver visibility at night and reduce the number of potential accidents, however they require high detection capability in dark environments to function. Improving the accuracy and reliability of such a system is an area that requires further research, many existing studies introduce methods that would greatly increase the computational complexity of such a system, something that must be avoided in a real-time system. This work explores existing techniques for the enhancement of object-detection in low light, and the lightweight methods that can be used to enhance detection. This project contributes a supplemental dataset for the training of object detection systems in low light, and proposes a new detection model YOLOv7-SE, based on the Squeeze and Excite attention module.

Keywords: Computer Vision, Automotive Platforms, Machine Learning

 Conference Details

Session: Poster Session A at Poster Stand 59

Location: Sir Stanley Clarke Auditorium at Tuesday 7th 13:30 – 17:00

Markers: Gary Tam, Arnold Beckmann

Course: BSc Computer Science FI, 3rd Year

Future Plans: I have a job lined-up