In the poultry industry, preventing chickens from developing illnesses during hot summer days is a significant concern. The early detection of abnormal behavior in chickens is crucial for farmers to promptly monitor their health status, and hence, research in this direction is paramount. In this paper, we proposed an automatic image recognition method, employing various software combinations, to analyze whether chickens exhibit abnormal mouth opening. The study began by obtaining images of chickens on the farm, capturing and observing any instances of mouth opening. The training process involved utilizing an image annotation tool, LabelImg, to mark instances of chickens and their mouth openings. Subsequently, the You Only Look Once (YOLO) algorithms, YOLOv3 and YOLOv4, were employed to train on images of chickens with open mouths. Once the YOLO training was completed, the Open Source Computer Vision Library (OpenCV) was used to read and access videos and images, enabling the observation of output results to determine the presence of mouth opening phenomena. In this approach, we integrated an advanced technology to automate the monitoring process, potentially providing farmers with timely alerts regarding any health issues in their chicken population. By leveraging computer vision and machine learning techniques, this method enhances efficiency and accuracy in detecting abnormal behavior, contributing to improved poultry health management strategies.