Unveiling the Hidden Patterns: Deep Learning Approach for Early Detection and Severity Classification of Sugarcane Brown Spot

Sugarcane brown spot (SBS) is a global fungal disease that threatens sugarcane crops. SBS treatment and yield preservation need early identification and precise severity categorization. We present a unique SBS detection and severity multi-classification method utilizing a convolutional neural network (CNN) Long-short term memory (LSTM) model. 20,000 sugarcane leaf images of various severities were gathered. Binary detection distinguished healthy and sick pictures, while severity multi-classification classified the disease into four severity levels. Our CNN-LSTM model accurately differentiated healthy and sick sugarcane leaf pictures at 98.11% in the binary detection phase. The algorithm accurately classified SBS severity levels during severity multi-classification. The model identified SBS severity with 93.87% accuracy, enabling focused illness treatment. CNN and LSTM architectures made our model work. CNN collected discriminative characteristics from sugarcane leaf pictures, whereas the LSTM network recorded temporal relationships, enabling disease progression analysis. The CNN-LSTM combination model can quickly detect diseases, assess severity, and optimize resource allocation in agriculture. This research advances automated SBS diagnosis and severity assessment, improving illness management.