Real-time Detection of Terminal Burn Defects Using YOLOv7 and TensorRT

Abstract

This study focuses on the detection of defects that may occur during the production process of electrical cable terminals. Cable terminals are critical for increasing the reliability of electrical connections and usually contain ends made of conductive metals and insulating materials that protect these ends. However, defects such as burns that may occur during the production phase can lead to malfunctions and performance decreases in electrical systems. YOLOv7 enabled effective defect detection, while TensorRT optimized performance for real-time processing. This algorithm was integrated with TensorRT on NVIDIA Jetson Nano hardware to achieve high detection speeds and accuracy rates, operating at 18 FPS (frames per second). The use of YOLOv7 enabled defects to be detected effectively, while TensorRT integration optimized system performance, allowing the defect detection process to be carried out in real time. This approach contributes to improving quality control processes on production lines and preventing potential malfunctions. ©2024 IEEE.