Sugarcane Leaf Disease Classification Using CNN–XGBoost Hybrid Model

Sugarcane is a vital commercial crop, and leafdiseases significantly impact its productivity andeconomic value. Timely and accurate identification ofthese diseases is essential for effective crop managementand yield improvement. This study presents a hybriddeep learning–based framework for sugarcane leafdisease detection by integrating Convolutional NeuralNetworks (CNNs) with the XGBoost classificationalgorithm. The CNN model is utilized as an automatedfeature extractor, learning high-level spatial and texturerepresentations from sugarcane leaf images that capturedisease-specific visual characteristics. These deepfeatures are subsequently provided as input to anXGBoost classifier, which leverages gradient boosting toachieve precise and robust disease classification. Theproposed hybrid CNN–XGBoost model is evaluatedusing a publicly available sugarcane leaf disease datasetand benchmarked against standalone CNNarchitectures and conventional machine learningclassifiers. Experimental results indicate that the hybridapproach delivers superior classification accuracy,enhanced robustness, and improved computationalefficiency compared to traditional methods. The combination of deep feature extraction and ensemble-based classification effectively mitigates overfitting while improving generalization performance. Thisresearch demonstrates that hybrid deep learning modelscan serve as a reliable and efficient solution forautomated plant disease diagnosis, contributing toprecision agriculture and sustainable crop managementpractices.