Detecting Disease In Livestock and Recommending Medicines using Machine Learning

Livestock is one of the critical socioeconomic assets in developing countries like India. However, the lack of a reliable and timely diagnosis system for identifying livestock diseases has led to significant losses in the livestock population, hindering efforts to achieve food security and reduce poverty in the country. To address this issue, a study proposed the integration of an expert system with machine learning and image processing. According to the 2019 Livestock Census, India has a total livestock population of around 535 million, which includes cattle, buffalo, sheep, goats, and pigs. The livestock sector is an important source of income for millions of households in India. As per the 2019 Livestock Census, there are around 145 million households involved in livestock farming and related activities in the country. The livestock sector contributes significantly to the Indian economy. In 2020, the total value of livestock output in India, excluding the value of horses, ponies, mules, donkeys, camels, and yak, was estimated to be around 9.37 trillion Indian rupees (around 125 billion US dollars), which accounts for around 4.2% of country's GDP. The dairy sector is a major contributor to the livestock economy in India. It accounts for around 70% of the total value of livestock output in the country. In addition to dairy, the livestock sector also provides meat, wool, leather, and other products. According to the latest available data from the Ministry of Statistics and Program Implementation, the contribution of the livestock sector to India's Gross Value Added (GVA) in 2020–21 was 5.04%. Since GVA is more accurate measure of the sector's contribution to the economy than GDP, it is concluded that livestock sector contributed around 5% to India's economy in 2020–21. According to a study by the National Dairy Development Board (NDDB) of India, there is an average of one veterinarian for every 5,000 cattle in India. This study improves the accuracy of livestock disease diagnosis and minimize losses by leveraging Convolutional Neural Networks (CNN) integrated with LSTM and ResNet50 algorithms. This model analyzes uploaded livestock images and accurately identifies specific diseases, providing prompt information on the disease name, recommended medicine, and food recommendations. By offering quick and accurate diagnoses, proposed model has the potential to significantly improve the livestock industry in India and contribute to poverty reduction efforts. To further enhance the performance, data augmentation techniques is proposed to increase dataset diversity, transfer learning with pre-trained models to leverage existing knowledge, ensemble learning to combine multiple models' predictions, continuous model refinement through regular updates and collaborations with domain experts, and user feedback and usability testing to improve the system's user experience. Implementing these strategies will contribute to the ongoing development and effectiveness of the livestock disease diagnosis system, facilitating better health management and reduced losses in the livestock sector.