Plant diseases are currently one of the most significant issues influencing modern agricultural production. One of the most important measures for monitoring, interpreting, and predicting how plant diseases may affect crop yield is the disease severity index. The essential plant disease treatment may be prescribed depending on the value of the disease severity, which is critical in minimizing additional yield loss. Traditional methods for diagnosing the severity of plant diseases often involve a skilled expert conducting visual analysis of specimens or plant tissue. However, this procedure is time-consuming, costly, and inefficient. As a result, now is the time to develop disease evaluation methods that will be beneficial in current agricultural production. In this study, YOLOv7 and YOLOv8 object detection methods were utilized to perform multi-class classification of various plant diseases using the PlantDoc dataset, with the goal of reducing losses caused by disease outbreaks. The YOLOv7 model demonstrated superior detection performance and exhibited strong generalization capabilities in complex scenes when evaluated on the PlantDoc dataset. It achieved high scores in metrics such as mean average precision (mAP), F1-score, precision, and recall of 60.99, 57.63, 52.83, and 63.38, respectively. The model’s effectiveness and efficiency are better than other object detection models.