Unsupervised Few Shot Image Classification

In the field of computer vision, unsupervised few-shot image classification refers to the task of training models using only unlabeled datasets during the pre-training or meta-training phase. The goal is to achieve rapid recognition and classification of new categories with limited labeled samples by learning the intrinsic structures and features from the unlabeled data. This task has significant practical value, as it can effectively reduce the cost of manual labeling and enhance the model's adaptability and generalization in real-world scenarios.