Zero-exemplar deep continual learning for crop disease recognition: a study of total variation attention regularization in vision transformers
With the increasing integration of AI technology in the food industry, deep learning has demonstrated its immense potential in the domain of plant disease image recognition.However, there remains a gap in research between models capable of continual learning of new diseases and addressing the inherent catastrophic forgetting issue in neural network