Gluten identification from food images using advanced deep learning and transfer learning methods

Area of Science:

• Computer Vision
• Artificial Intelligence
• Dietary Management

Background:

• Food image recognition is crucial for dietary management, especially for individuals with dietary restrictions.
• Accurate identification of gluten-containing foods is essential for celiac disease management.

Purpose of the Study:

• To develop a novel approach for gluten image classification using Convolutional Neural Network (CNN) architecture.
• To assist individuals with celiac disease in identifying gluten-containing foods through accurate food image recognition.

Main Methods:

• Utilized a curated dataset of 20,000 food images from the Food101 dataset, focusing on common recipes.
• Employed a Convolutional Neural Network (CNN) architecture, specifically leveraging an EfficientNet pretrained model for image classification.
• Ensured high-quality training and testing through meticulous data labeling.

Main Results:

• Achieved a training accuracy of 99.02% and a validation accuracy of 98.38% using the EfficientNet pretrained model.
• The model demonstrated high effectiveness, obtaining an accuracy of 99% on a test set of 2000 images.
• Results highlight the model's capability in accurate gluten classification.

Conclusions:

• The developed food image recognition model is effective for gluten classification.
• This tool offers significant potential utility for celiac patients in managing their dietary intake.
• The study contributes a valuable resource to the field of food image recognition for dietary management.