Pendeteksian Sepeda Motor di Jalur Khusus Sepeda Menggunakan Algoritma Pendeteksi Objek YOLO
Bicycles as transportation by the public is increasing. The government supports this by providing bike lanes. However, limited supervision by security forces poses a safety threat for cyclists against irresponsible persons. This research proposes a method of detecting vehicles in bike lanes from CCTV video using YOLOv3 algorithm and creating a detection area on the bike lane. The YOLO algorithm was chosen because it has a high FPS value. Vehicles that can be identified are motorcycles and bicycles. The bike lanes detection area is determined manually following traffic markings, while the vehicle object detection used YOLO v3 Tiny. YOLO v3 Tiny obtained an mAP accuracy value of 72.73% with CPU processing performance reaching three frames per second. Based on application testing, the application can distinguish bicycles and motorcycles in the bike lanes. However, YOLOv3 Tiny's low mAP accuracy can cause misdetection in applications.