Enhancing Disease Impact Analysis Through Deep Learning Methods for Temperature–Based Crop Disease Detection

Through the utilization of thermal images data, this research endeavors to evaluate the possible uses of deep learning (DL) in the identification and diagnosis of crop illnesses. When it comes to capturing the complex and spatially dependent fluctuations that are associated with the course of disease, traditional methodologies frequently depend on single point temperatures, which may prove to be insufficient. Through the use of a dataset consisting of thermal images that have been labeled for both healthy and pathological states, we present a unique image-based deep learning model. With the help of this model, the complete thermal image is analyzed, and complex details that go beyond conventional temperature averages are extracted. A 92.98% accuracy rate is achieved by our model, which exhibits greater performance in comparison to other methods that are currently in use. Furthermore, the model demonstrates a high level of precision (93.38 percent) and area under the curve (99.46 percent), which indicates that it is able to differentiate among healthy and unhealthy crops. Through the visualization of the model's predictions accompanying thermal images, we further investigate the interpretability of the information. Because of this analysis, certain thermal patterns that are related particular disease have been revealed, which may make it possible to implement specific remedies based on individual thermal signatures. A significant step forward in the detection of crop diseases through the use of DL is represented by this work. The integration of real-time sensors for temperature, weather data, and multispectral photography is one of the future topics of research that will be developed in order to produce a comprehensive picture of crop health.