FreshnessAI: AI-Based Freshness and Shelf-Life Prediction for Cut Fruits

Cut fruits are minimally processed products that deteriorate quickly after cutting, and shops and food businesses usually judge their freshness by eye. Such inspection is subjective and gives no reliable estimate of how long a piece of cut fruit can still be sold or served safely. This paper presents FreshnessAI, a proposed web and mobile system that applies computer vision and deep learning to assess the freshness of cut fruits from images and to provide shelf-life (expiry) guidance. The system is built around a Python Django backend with two clients: a web application for administrators, fruit shops and restaurants, and a Flutter-based Android application for customers. An AI inference engine built with PyTorch and YOLO pre-processes an uploaded image, detects the fruit type and location, assigns a freshness status (fresh or spoiled) with a confidence score, and produces a shelf-life estimate with advice. Results and images are stored in a MySQL database and media storage, and the YOLO model is trained offline through a pipeline of dataset collection, annotation, augmentation, splitting and GPU-based training. We survey twenty research papers, identify the gaps they leave for cut fruits, and present the requirements, architecture and data flow design. Dataset preparation, training and evaluation are planned for the next phase, and no performance figures are claimed in this paper