Semi Supervised Image Classification Cold

Semi-Supervised Image Classification (Cold Start) is a subtask in the field of computer vision, focusing on intelligently selecting the most informative samples from a large set of unlabeled images for annotation to build an efficient model. This task avoids the class-stratified random sampling method commonly used in traditional semi-supervised learning, requiring the initiation of labeling without relying on class information. As such, it enhances the model's generalization and robustness, making it highly valuable for practical applications.