Automated Non-Contact Temperature Measurement System for Truck Tires and Brakes Using Thermal Imaging and YOLO

During truck operation, the temperature of tires and brakes can rise significantly, potentially leading to brake system failures. In this work, we present an innovative approach to automotive safety inspection, focusing on the critical components of trucks: tires and brakes. Our work aims to design and develop a non-contact temperature measurement system for truck tires and brakes, intended for installation at Heavy Vehicle Safety Stations. Since the tires and brakes are located in difficult-to-reach areas, so that the existing temperature measurement methods used by inspectors are inefficient. We propose an automatic non-contact temperature measurement system utilizing a thermal camera and an HD camera installed on the ground. To achieve optimal viewing angles, both cameras are installed on a motorized platform with two degrees of freedom, allowing rotation along the yaw and pitch axes. A YOLO algorithm, trained with a customized dataset, detects the position of tires and brakes from RGB images, while temperature is extracted from corresponding pixel positions in the thermal camera's output. To reduce the computational burden of YOLO, we utilize a Kernelized Correlation Filter (KCF) as a tracking algorithm. The system can accurately detect tires and brakes in testing data set, achieving an F1-score of 95.3% and an Intersection over Union (IoU) of 0.79. Furthermore, by comparing computation times, the processing time with tracking was reduced to approximately 0.165 seconds per frame, down from about 1.08 seconds per frame without tracking.