Finding Vacant Seats in a Public Seating Area with Object Detection using YOLOv10

This paper presents an approach to real-time detection of vacant seats in public seating areas using YOLOv10. In places where finding available seating can be slow and inefficient, this system offers a solution by utilizing YOLOv10 to find available seats by analyzing a live surveillance camera footage of the seating area. The system processes the video to pinpoint vacant seats and indicate their locations on a user interface using bounding boxes. Additionally, it counts the number of currently unoccupied seats, providing instant information on seating availability to the user. By automating the detection and monitoring of vacant seats, the system reduces the manual effort involved in this task and improves overall user experience in public spaces. The system is developed in Python, and it utilizes OpenCV to seamlessly connect with surveillance cameras. This approach not only has the potential to assist in crowd management but also demonstrates the practical application of computer vision techniques in real-world scenarios. By training YOLOv10 using a dataset containing 500 images of an audience seating area, the model has fulfilled its main purpose, which is to locate available seats, mark them, and count their amount. This model achieved a strong mAP50 score of 0.95, along with a precision, recall, and F1 score of 0.95, 0.92, and 0.94 respectively, indicating good overall accuracy.