Real-Time Mastitis Detection in Livestock using Deep Learning and Machine Learning Leveraging Edge Devices

• Published in: 2023 IEEE 17th International Symposium on Medical Information and Communication Technology (ISMICT) pp. 01 - 06
• Main Authors: Kumar Ghosh, Kawshik, Ul Islam, Md. Fahim, Efaz, Abrar Ahsan, Chakrabarty, Amitabha, Hossain, Shahriar
• Format: Conference Proceeding
• Language: English
• Published: IEEE 10.05.2023
• Subjects: Agriculture; Cows; Deep learning; Developing countries; edge devices; Image edge detection; livestock; machine learning; mastitis; Productivity; Real-time systems
• ISSN: 2326-8301
• DOI: 10.1109/ISMICT58261.2023.10152110

Livestock production is a crucial part of the global economy with a worth of estimated 1.4 trillion. It provides livelihoods for 1.3 billion people and supports 600 million poor rural household farmers in developing countries. In Bangladesh, it contributes 6.5% to the country's GDP. However, this industry faces substantial financial setbacks when contagious diseases transmit among their livestock. One of the most common and expensive diseases affecting the livestock industry is Bovine Mastitis. This paper presents a real-time system for detecting bovine mastitis in livestock using deep learning (dl) and machine learning (ml) techniques. The system aims to provide a timely and accurate diagnosis of mastitis, ultimately reducing costs and improving the efficiency of treatment. By utilizing dl and ml techniques, the system is able to analyze data collected from edge devices and make accurate predictions about the presence of mastitis. The dataset that has been used for the classification contains both an Image dataset consisting of 1341 images and a Numerical dataset that had been taken from 1100 cows over a period of six days. The edge device utilizes sensors and cameras to collect data from the cow, which is then processed through ml and dl algorithms using Raspberry Pi and cloud computing respectively, and then displays if the cow is infected with mastitis or not. Inception V3 and RandomForest algorithms were used for dl and ml, respectively, and had an accuracy of 99.34% and 99% respectively. The proposed system has the potential to significantly reduce the economic impact of this disease in the dairy industry of Bangladesh and other developing countries by providing timely and accurate diagnosis and helping to improve treatment efficiency and protect the health and productivity of livestock animals.