ZUT-FIR-ADAS

Pedestrian detection has never been an easy task for computer vision and automotive industry. Systems like the advanced driver assistance system (ADAS) highly rely on far infrared (FIR) data captured to detect pedestrians at nighttime. The recent development of deep learning-based detectors has proven the excellent results of pedestrian detection in perfect weather conditions. However, it is still unknown what is the performance in adverse weather conditions. In this paper, it is introduced a 16bit thermal data dataset called ZUT (Zachodniopomorski Uniwersytet Technologiczny) having the most extensive variety of fine-grained annotated images captured in 4 biggest European Union countries captured during severe weather conditions. In addition to this, we also provide a synchronized Controller Area Network (CAN) bus data, including driving speed, brake pedal status, and outside temperature for future ADAS system development. Furthermore, we have tested and provided 16-bit depth modifications for Yolov3 deep neural network (DNN) based detector reaching mean Average Precision (mAP) up to 89.1%.