Poultry-Diseases Recognition System with Deep Learning

Poultry farming is an important part of agriculture that provides food and income to many people. But poultry birds often suffer from diseases like Coccidiosis, Salmonella, and Newcastle Disease. These diseases spread quickly and can cause big losses for farmers. Traditional methods to detect these diseases, like lab tests, are slow, expensive, and not available in rural areas. So, there is a need for a fast and affordable way to detect diseases early. In this project, we used deep learning techniques to build a model that can detect poultry diseases from images of chicken feces. We used CNN models like Xception, EfficientNetB0, and EfficientNetB3 with transfer learning. The models were trained using a large dataset of over 500,000 images divided into four classes: Healthy, Coccidiosis, Newcastle Disease, and Salmonella. After testing all models, Xception gave the best results with an accuracy of 92.78% and balanced performance across all classes.­­­­ This system can be used in real-time through mobile apps or IoT devices. It helps farmers and veterinarians detect diseases quickly, even in remote areas. This not only reduces the cost and time for testing but also helps control the spread of diseases early. In the future, this system can be improved further by adding more disease types, real-time alerts, and explanations for its prediction