Meat Freshness Classifier with Machine and AI

• Main Authors: Bhuiyan, Zarif Wasif, Redwanul Haider, Syed Ali, Haque, Adiba, Hasan, Mahady, Uddin, Mohammad Rejwan
• Format: Conference Proceeding
• Language: English
• Subjects: Arduino UNO Microcontroller; Cameras; Feature extraction; IoT; Machine learning; Machine learning algorithms; Meat Freshness; Microcontrollers; MQ135 gas sensor; MQ4 methane natural gas sensor; Supply chains; Visualization
• ISSN: 2642-6102
• DOI: 10.1109/TENSYMP55890.2023.10223681

Using machine learning and artificial intelligence techniques, this thesis presents a novel approach to detecting meat freshness. The proposed system consists of two gas sensors MQ135 and MQ4 to capture the odors emitted by the meat samples, an ESP32-CAM, and an Arduino UNO microcontroller to process the sensor data and extract relevant features. A machine learning model is trained using a dataset of labeled meat samples with known freshness levels. The proposed technique accurately categorizes the freshness of meat samples with a classification accuracy of over 90%, showing the potential of machine learning and artificial intelligence in improving the precision and effectiveness of this procedure. The technology is transportable and compatible with current meat processing equipment. This gives the food business a dependable, automated method to raise the security and caliber of meat goods. Overall, the study's findings show that the suggested system is a reliable way to classify the freshness of meat. This project proposes a novel approach to detect meat freshness using two gas sensors along with a camera that employs image processing AI techniques to overcome challenges posed by added color in meat. Although there were some limitations regarding Data Availability, Subjectivity of freshness Determination and many other real-time assessments. Despite the limitations the ML and AI can help to mitigate some of the limitations and improve overall performance.