Pengenalan Jenis Masakan Melalui Gambar Mengunakan YOLO

Alexander William Sutjiadi (Program Studi Informatika)
Kartika Gunadi (Program Studi Informatika)
Leo Willyanto Santoso (Program Studi Informatika)

Jurnal Infra
Vol 9, No 2 (2021)

Most input systems in food calorie information applications still use manual input, where the user needs to write the name of the food. This can cause difficulties for the user who does not know the name of the food. But with the development of technology in object detection, this can be an easier and faster process by using images to recognize the type of food.This research uses You Only Look Once method to recognize the type of food from the input images. The type of YOLO model used is the YOLOv3 model which has a modified convolution layer with Xception model that has high accuracy and detection speed.The result obtained is that the modified YOLO model has higher accuracy and faster detection speed than the standard YOLOv3 model. The Highest mAP result achieved was 85.13% with 0.088742 seconds average detection time.