Fruits Calorie Prediction Using Machine Learning

Abstract

The proposed method in this work provides a unique solution through image-based analysis for predicting fruit calorie content using object detection and segmentation with the YOLOv8 model. The model was trained on a custom dataset comprising diverse fruit images to detect and seg-ment seven types of fruits. From the segmented fruits, we computed the relative area of each fruit, which served as a predictor in estimating calories using a linear regression model. The goal was to create an effective, reliable food-image-based calorie estimation system. The YOLOv8 model performed exceptionally well in fruit detection, achieving precision and re- call almost equal to 1.00 for all categories except for oranges, where the F1-score was slightly reduced to 0.99 due to a high missed ratio of approximately 2%. The model's fruit detection and classification tests demonstrated an accuracy of 1.00, reflecting its robustness in identifying fruits.The linear regression model achieved an R2 = 0.987 for calorie estimation, showing a high corre- lation between the fruit area and calorie content. The model captured 98.7% of the variance I calorie values, demonstrating its ability for accurate predictions based on the segmented fruit ar-eas. Major outcomes of this study include high accuracy and the ability to detect and segment fruits using YOLOv8. The fusion of improved object detection techniques with simple machine learning models for nutrition estimation proved to be effective. Overall, this system provides a calorie estimation tool that can assist users in mobile health applications, smart kitchen appliances, andietary consulting services. Future work includes expanding the range of fruits in our dataset anexploring more sophisticated machine learning models to enhance calorie prediction accuracy However, the generalizability of this study may be affected by restrictions in terms of fruit typeand lighting conditions from which the dataset was prepared. Users (Clients): Individuals or businesses seeking office rentals can browse and book office based on their specific requirements, such as office size (e.g., square footage), location preferences, and included facilities. Employees: Responsible for managing client interactions, employees utilize video conferencing to provide detailed office tours, facilitate product sales, and ensure seamless operations of the rental services.