Combining Autoencoder and Yolov6 Model for Classification and Disease Detection in Chickens.

Among the causes of reduced production is a chicken disease, which can negatively affect consumer health. With the advancement of computer vision technology and profound innovations in the field of research, it has become increasingly important to analyze disease images collected by sensors in chickens to analyze the possibility of infection conveniently and efficiently. Consequently, research proposes to identify lesions using the Autoencoder and Yolov6 model to classify and detect diseases in chicken flocks. This model is suitable for different chicken breeds from many countries and regions. This method helps improve and enhance image recognition accuracy by incorporating the data enhancement method in the data preprocessing step. The results show that the value of val/mAP (average accuracy) obtained by the method proposed in this paper is 99.15%. Moreover, hit over 90% on the test dataset. This method can be applied to the early detection of disease-carrying chickens in the captive population, ensuring a quality food source for humans.