Precise and efficient disease detection in crops is highly significant in the pursuit of increasing agricultural productivity and sustainability. A hybrid deep learning model that uses ResNet50 and Gated Recurrent Units (GRU) is introduced in this study for precise classification of rice leaf diseases, which are one of the most severe issues in global rice cultivation. In this study, we fine-tuned our model using the “Dhan-Shomadhan” dataset which contains images of five different rice diseases: Brown Spot, Leaf Scaled, Rice Blast, Rice Tungro, and Sheath Blight. From these images, the spatial features are excellently extracted by the ResNet50 component while GRU enhances its ability to recognize temporal patterns and subtle disease traits over sequences. The proposed, Res-GRU hybrid model also achieved an amazing accuracy rate of 99.07% as well as high precision, recall, F1, sensitivity, and specificity score measures which demonstrate its robustness in classifying diseases. These outcomes showed how integrating deep learning technology into agriculture can provide a scalable automated solution that minimizes reliance on human efforts for inspections. This study also underscores the potential of integrating deep learning technology like GRU in image processing and into agriculture, offering a robust, scalable, and automated solution, that contributes to global food security.