Applying YOLO Model in iCAR for Multi-Detection: Object Detection and Hand Gesture Recognition

Computer vision technology is increasingly pivotal across various industries, particularly in transportation, where it plays a critical role in enhancing both efficiency and safety, especially in autonomous vehicles. This study addresses a significant challenge encountered by autonomous cars, exemplified by iCAR's inability to identify and provide access to traffic officers. The project's objective is to equip iCAR with the capability to recognize traffic officers and enable hand gesture-based access control. By leveraging Convolutional Neural Networks (CNNs) integrated with the You Only Look Once (YOLO V8) model, a robust model was developed and trained on a dataset comprising approximately 5000 images that include traffic officers and hand gestures. The study yielded promising results, as indicated by the F1 score analysis, with an average F1 score across all classes reaching 0.96, underscoring the model's effectiveness. Through the integration of computer vision technology, iCAR facilitates seamless human-vehicle interaction by accurately discerning traffic officers and granting them access via hand gestures.