APPLICATION OF THE YOLO ALGORITHM FOR HUMAN DETECTION IN A ROBOT RECEPTIONIST

Abstract

The receptionist robot is designed to enhance interaction with users in various environments, such as hotels, office buildings, and shopping centers. One of the primary challenges in the development of such robots is the fast and accurate detection of humans to enable real-time operation. This study applies the You Only Look Once (YOLO) method, specifically YOLOv5, for human object detection in receptionist robots. YOLOv5 was chosen due to its advantages in detection speed and accuracy. The research stages include system design, data acquisition, image processing, model implementation, and performance analysis using metrics such as mean Average Precision (mAP) and frames per second (FPS). Based on the results, it can be concluded that the YOLOv5 method is capable of detecting human objects with high accuracy and fast inference time. The model demonstrated optimal performance under bright lighting conditions, achieving a precision of 95% and a recall of 92%, with an inference speed of 30 FPS. Although there was a decrease in accuracy of approximately 4% under low-light conditions, the model remained consistent, achieving a precision of 88% and a recall of 85%. Additionally, the YOLOv5 model was tested under various scenarios, including changes in lighting, object distance, and crowd density within the frame. The results showed a slight decline in accuracy when objects were more than 5 meters away or when there were more than three people in a single frame, which led to an increase in false positives. Architecturally, YOLOv5 divides the input image into a 7x7 grid, where each grid cell predicts class probabilities and bounding boxes, enabling efficient and accurate detection. Therefore, YOLOv5 proves to be an effective solution for human object detection under various environmental conditions, although there remains room for improvement in specific scenarios such as low-light environments and high-density crowds.