Real-Time Detection of Fruits and Vegetables with Calorie Estimation Using Deep Learning

This paper presents a comprehensiveapproach to real-time fruit and vegetable detectionintegrated with calorie estimation using deep learningtechniques. With the rise of health awareness and the needfor accurate dietary assessment tools, the proposedsystem addresses a practical challenge by enablingautomated nutritional analysis through computer vision.A custom YOLOv5 object detection model was trained ona robust dataset containing 32 classes of commonlyconsumed fruits and vegetables. The model demonstrateshigh accuracy in diverse environments, includingvariable lighting and background noise, making itsuitable for real-world deployment. Detected items aremapped to a locally maintained calorie dictionary basedon standardized nutritional data, enabling calorieestimation per item and cumulative calculation formultiple items in a single frame. The system supports bothimage uploads and real-time webcam input and isdeployed as a lightweight web application built usingFlask. It offers fast inference, requiring no external APIs,ensuring data privacy and offline usability. The solutionhas potential applications in fitness tracking, dietplanning, educational tools, and mobile healthcaresolutions. Experimental results demonstrate strongperformance in terms of detection accuracy and responsetime, making the system a reliable aid for smart foodanalysis and calorie tracking.