Improving Vehicle Assistance Systems: Evaluation of Augmented Capabilities through Infrared Thermal Camera Integration

Abstract

Nighttime driving is difficult owing to low visibility and lights. Nighttime accidents are more dangerous due to reduced obstacle detection, poor vision, and trouble evaluating distances. Knowing the causes and dynamics of nighttime accidents is essential for improving road safety and preventing collisions when natural light is limited. This study proposes using an infrared thermal sensor to assist drivers in mitigating the issue of inadequate light at night, with the ultimate goal of preventing crashes under such circumstances. The investigation compared the infrared thermal camera sensor with the normal camera visual to evaluate how well it worked at night. The testing has been done on the road in Pekan, Pahang. Yolov8 deep learning has been integrated with both cameras to detect items like cars, motorcycles, and traffic lights. The test findings demonstrated how temperature variations can be utilized to precisely detect items on different types of roadways. The study showed that infrared thermal sensors are impressive at detecting traffic lamps, motorcycles, and vehicles. The infrared camera's actual detection on confusion matrices was 0.98 for traffic lamps and 0.87 for motorcycles and vehicles. This shows how well the infrared thermal camera works in dark conditions, with a faster frame rate of 64.94 fps than regular cameras at 55.25 fps. The results of this study demonstrate that using infrared technology can enhance object detection capabilities and, hence, enhance nighttime road safety.