Attention-enhanced Vision Transformer for Tea Leaf Disease Classification: An Experimental Study on an Assam Tea Leaf Dataset
It is suggested that lightweight feature recalibration can enhance a compact ViT without substantial parameter overhead, however, the limited dataset size and absence of an independent external test set require cautious interpretation and further validation across diverse field conditions.