VGG16-AttnNet: a banana leaf lesion recognition model combining VGG16 and self-attention mechanism

As one of the essential fruits worldwide, bananas have leaf diseases that seriously affect yield and quality. Therefore, a model that can accurately recognize these lesions is crucial for improving agricultural productivity and banana quality. This study, VGG16 was used as the infrastructure with a multi-layer self-attention mechanism layer to train, validate, construct, and train a model to distinguish between healthy leaves and different lesions using a banana leaf lesion image dataset. The experimental results show that the model with the superimposed attention mechanism layer gives better results of 92% compared to the original VGG16 model.