Technical Paper
Sensor Fused Low Light Pedestrian Detection System with Transfer Learning
2024-04-09
2024-01-2043
Objection detection using a camera sensor is essential in implementing Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD). Due to the recent advancement in deep Convolution Neural Networks (CNNs), object detection during daytime has achieved state-of-the-art performance. However, using an RGB camera alone, the object detection system can dramatically drop the system's performance under poor lighting conditions such as sun flare, snow, fog, and nighttime. That is because the camera sensors are susceptible to lighting conditions. The nighttime vision systems are relatively less studied compared to the daytime system. For a nighttime pedestrian detection system, sensors less sensitive to lighting conditions are necessary to detect pedestrians accurately with low lighting conditions. This paper proposes a sensor-fused nighttime pedestrian detection system.