Analysis of the Efficiency of Computer Vision for the Detection of Vehicles and Pedestrians in Traffic
: Within this research, the focus was on analyzing the effectiveness of Computer Vision (CV) in detecting vehicles and pedestrians in traffic. The YOLOv5 model was utilized for object detection, along with publicly available, unmodified libraries like OpenCV and TensorFlow. The approach involved a careful selection of three different traffic scenarios: a rainy day, daytime, and night-time, with the intention of creating realistic conditions for testing the performance of vehicle and pedestrian detection systems. An algorithm for detecting pedestrians and vehicles was implemented, contributing further to road safety. Through experiments, exploration was conducted into how various factors, such as weather conditions and lighting, influence the accuracy of the system. Following a meticulous analysis of the results, situations in which the system exhibits high detection accuracy, as well as those that pose a challenge to the system were identified, in order to provide a profound understanding of different aspects of pedestrian tracking and vehicle detection. Through the application of image analysis techniques, the focus was on identification of key features of pedestrian crossings, contributing to the recognition of potentially dangerous situations. The objective was to draw accurate conclusions regarding the system's performance under actual traffic conditions, thus enhancing the overall comprehension of how these technologies effectively contribute to improving road safety.