JMIR Publications

ABSTRACT

Background:

In recent years, the field of computer vision has seen remarkable advancements, particularly with the rise of deep learning techniques, in identifying objects in images. We investigate the application of object detection models, specifically You Only Look Once (YOLO), in the context of food and portion recognition.

Objective:

The objective of the study is to assess the performance of three YOLO models in accurately identifying food components within images and their proportions in terms of the Swedish plate model as per the guidelines of the National Food Agency.

Methods:

The research utilizes a custom dataset comprising 3707 images with 42 different food classes. Various preprocessing and augmentation techniques were applied to enhance dataset quality and model robustness. The performance of the three YOLO models (YOLOv7, YOLOv8, and YOLOv9) are evaluated using precision, recall, mean Average Precision (mAP), and F1 score metrics.

Results:

Our analysis demonstrates that YOLOv8 showed higher performance. While the metrics were satisfactory, the models still had difficulty distinguishing between different food items.

Conclusions:

We determine that further enhancements to both the technology and the training are required for such tools to be included in health monitoring applications.