Automated Detection of Rotten Fruits and Vegetables using Deep Learning

Fruit and vegetable spoilage is a universal issue, causing significant loss to food wastage, economy loss, and health risk. Traditional quality control methods are human intervention oriented, which lacks uniformity and practicability at an industrial level. In the present paper, the authors propose MobileNetV2, a lean but powerful convolutional neural network (CNN), to fulfill AI-based fruit and vegetable spoilage classification as from healthy to rotten. The system contains an end-to-end pipeline consisting of image preprocessing, fast real-time model inference using FastAPI, and a friendly frontend made using Streamlit. The application also includes storage instructions and safety protocols and integrates an AI chatbot to support users. Experimental results demonstrate superior accuracy (94.8%), low latency (<1.2 seconds), and superior generalization across diverse produce. This research emphasizes the potential of deep learning in minimization of wastes and ensuring food safety across supply chains.