Rice, which is extensively utilized as a staple food across the globe, assumes significant importance in the quest for superior rice products. However, the incidence of rice illnesses may prevent rice-based goods from being produced at their best and of the highest quality. Precision agriculture investigates the development of automated disease detection and categorization systems in great detail. Because rice is grown in vast, moist regions, it might be difficult to detect these infections. A Deep Learning-based multiclass paddy disease finding model (DL-MPDP) is presented in this research, for accurate identification and categorization of afflicted areas in paddy plants. IoT cameras are utilized to acquire unprocessed images of paddy fields. The images undergo preprocessing techniques such as LANCZOS, CLAHE, and Wavelet to increase their quality before additional analysis. Then, the pre-processed image can be subjected to Feature Extraction (Texture Feature, Shape Feature, and Edge Feature) via Onehot Encoding Technique. Then, the dimensions of the extracted image will be reduced via a new Kernel-based Principal Component Analysis (KPCA). Subsequently, from the dimensionally reduced data, the optimal features will be provided via the newly modified Flower Pollination Algorithm (mFPA). The Recurrent Neural Network with Long Short-Term Memory (HRNNLSTM) model that makes the final detection (presence or absence of disease) is trained using the selected optimal features. Moreover, to enhance the detection accuracy, the weight function of HRNNLSTM is finetuned using the Crayfish Optimization Algorithm.