Overview
It is an AI-powered mobile application capable of recognizing multiple food items on a cafeteria tray from a single photo and automatically calculating the total calories and price. A suitable dataset was created for this task, and all images were thoroughly annotated using Roboflow. Each dish was linked to a JSON-based information structure containing its name, calorie value, and price.
The dataset was divided into 70% training, 20% validation, and 10% test splits. The trained YOLOv8 model was optimized to accurately detect the foods on a tray. The model was integrated into a REST API using Python and FastAPI, and the Flutter-based mobile application was designed to deliver real-time results through this API.
Key Features
- Real-time tray and food detection using YOLOv8
- Automatic calorie calculation for each detected meal
- Total tray price calculation and menu-based discount / savings estimation
- Flutter-based Android mobile application with a simple 3-page flow
- FastAPI backend that processes uploaded images and returns JSON results
Gallery
My Contribution
- I contributed to the annotation of the entire dataset.
- I assisted in the training and development of the YOLOv8 model.
- I contributed to backend integration and API development using Python and FastAPI.