Comparative Analysis of YOLOv8 and SSD for Real-Time Object Detection in Driving Environments

Real-time object detection is crucial factor for various intelligent transportation systems, surveillance systems, and the self-driving car technologies. This research aims to evaluate the performance of two most commonly used object detection algorithms, YOLOv8 and Single Shot Multi Box Detector (SSD), with the help of the BDD100K dataset. Both algorithms are trained and evaluated under the same experimental settings to compare their performance effectively. Various performance evaluation metrics were used to compare the performance of the YOLOv8 and SSD algorithms. These metrics include precision, recall, F1-score, mean Average Precision (mAP), and frames per second (FPS). Based on the experimental results, it concluded that the YOLOv8 algorithm performs better with respect to accuracy and robustness for detecting objects in a complex traffic environment, especially for small and overlapping objects. Though the SSD algorithm performs faster training with a lower computational cost, the accuracy of the object detection results is relatively lower compared to the YOLOv8 algorithm.