An Integrated Approach to Spoiled Food Detection in IoT: Combining Gas Sensor Technology and Machine Learning

This research presents a hybrid solution for the timely and accurate detection of spoiled food in an Internet of Things (IoT) environment. The proposed system integrates hardware-based sensing with a machine learning (ML) model to enhance the efficiency and reliability of food quality assessment. The hardware component utilizes an MQ3 gas sensor interfaced with a Raspberry Pi, designed to detect specific gases emitted by spoiled food items. The real-time gas concentration data is transmitted to the ThingSpeak server, providing a continuous monitoring system for potential spoilage indicators. In parallel, a robust ML model is developed using a diverse dataset of annotated images featuring both spoiled and fresh fruits. The model, trained on a convolutional neural network (CNN), is adept at classifying the visual attributes associated with food spoilage. The ML model is implemented for on-device processing, enabling rapid and accurate assessment of food quality based on images captured by an embedded camera.