Dragon-fruit Stem Health Classification with Deep Learning and Attention Mechanisms
DOI:
https://doi.org/10.70917/ijcisim-2026-2658Keywords:
Plant disease classification, Hybrid Deep Learning Model, ResNet50, EfficientNetB0, attention mechanism, Grad-CAM, interpretability, classification accuracy, early disease detection, robustnessAbstract
Plant disease detection is important for maintaining the health and quality of crop yields. For the detection of health
problems in images of dragon-fruit (Hylocereus) stems, we present a deep learning architecture that is augmented by the
application of spatial, channel, and domain-specific attention mechanisms. For the purpose of improving accuracy and robustness,
the model architecture is combined using features from ResNet50 and EfficientNetB0 backbones, along with separate attention
branches. Training and testing were performed using a proprietary dataset of images of dragon-fruit stems that are healthy as well
as showing various levels of disease. The model initially used frozen feature extractors for training, which were subsequently finetuned to improve overall performance. Experimental results provide high accuracy for classification, with ROC-AUC values above
0.94 for all classes. The proposed method enables precision agriculture operations by providing a robust method for early detection
of dragon-fruit diseases. Future work will include adapting the model to real-world field images and further optimising it for
application in agricultural monitoring systems
