Chicken Diseases Detection and Classification Based on Fecal Images Using EfficientNetB7 Model

Abstract

The agriculture sector, particularly the chicken and poultry industries, is under pressure to produce more due to the increased demand for livestock products among consumers. Increased poultry production can lead to the enhanced spread of many infectious diseases in chickens, which can result in high bird fatality rates and significant financial losses. A shortage of trustworthy professionals or delayed diagnosis cause farmers to be losing an extensive number of domestic chickens. Deep learning algorithms can help the early detection of illnesses. This paper proposes a system based on convolutional neural networks to categorize chicken illnesses by identifying healthy and harmful fecal images. Unhealthy images may indicate a poultry illness. Through the use of deep learning algorithms and image analysis of chicken feces, the most common illnesses that affect chickens may be rapidly identified. With the use of a convolutional neural network (CNN) architecture, this research developed a model to identify different chicken ailments by classifying fecal images into two groups: those representing symptoms associated with healthy conditions and those representing symptoms associated with potentially dangerous conditions like Newcastle diseases, Coccidiosis, or Salmonella. To determine if chicken feces fell into one of four categories with the least amount of loss utilized the EfficientNetB7 model with additional layers that extracted the most appropriate features from the fecal images and achieved the highest accuracy. With an accuracy of 97.07%, the new proposed model generated the greatest results when compared to the aforementioned models.