AI-Based Poultry Disease Detection Using Fecal Image Classification with EfficientNetV2

In this paper, an AI-based poultry disease detection system is presented using fecal image classification. Four poultry health conditions, such as Coccidiosis, Healthy, Newcastle Disease, and Salmonella, are classified using the transfer learning method with EfficientNetV2B0. In the experiment, the model achieved an approximate 98 percent accuracy with high precision and recall values for the classes. The proposed system is beneficial to the farming community as it is cost-effective and allows the detection of poultry diseases in their early stages. Farmers can use the proposed system to diagnose their poultry diseases by uploading the images of their feces into the proposed system.