A Hybrid RNN Model for Accurate Paddy Disease Diagnosis

There are issues with standard RNNs, like not remembering long sequences, complex structure, and lower speed. To overcome these issues, RNN is modified by introducing LSTM and GRU, which provide a simpler architecture using GRU, increase the speed by GRU, and allow for remembering long sequences by LSTM. This modified RNN is applied over paddy crop diseases, where disease identification, progression, or minimization is detected in the time-series images captured for disease prediction analysis. Most of the southern states of India are dependent on rice or paddy for their daily food. To increase crop yield and revenue to the paddy farmers, identify the diseases using modified RNN with the help of LSTM and GRU, and alert the farmers to take action like applying the right amount of pesticide portion over the crop, crop cutting, and other scenarios. If any unnecessary phases and redundancies are present, they will be removed using the proposed approach. This combined approach helps to detect paddy crop diseases very early and converts them into healthy crops by taking appropriate steps. Later, after a time frame, the detection of the disease is again present, or it is removed from the crop and analyzed using a modified RNN. The metrics used here to evaluate the reliability of processing are accuracy and performance.