Sistem Pendeteksi Ketersediaan Tempat Duduk pada Perpustakaan berbasis Computer Vision Studi Kasus : Perpustakaan Unand Ruangan The Gade Creative Lounge

Abstract

Libraries as information centers often face challenges in seating management, especially when visitor numbers are high. This can cause difficulties for new patrons in finding empty seats, as well as disrupting other patrons' activities. Existing solutions, such as reservation systems and sensors, are often less effective and face accuracy issues. This research proposes a computer vision-based empty seat detection system to improve space efficiency. Using Roboflow 3.0 object detection model trained on Roboflow GPU and implemented on Raspberry Pi 4B with webcam, this system detects and displays empty seats through website with room mapping. The model demonstrated performance with Mean Average Precision (mAP) 68.4%, Precision 70.4%, and Recall 68.7%, showing effectiveness in detection although there is still potential for accuracy improvement.