方法对比

	自监督语义分割	自监督视觉Transformer
领域	深度学习	深度学习
方法族	Machine learning	Machine learning
起源年份	2020–2022	2021–2022
提出者	Multiple groups (Caron et al.; Hamilton et al. among key contributors)	Caron et al. (DINO); He et al. (MAE)
类型	Self-supervised dense prediction	Self-supervised pre-training for vision transformers
开创性文献	Caron, M., Touvron, H., Misra, I., Jegou, H., Mairal, J., Bojanowski, P., & Joulin, A. (2021). Emerging Properties in Self-Supervised Vision Transformers. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 9650–9660.	Caron, M., Touvron, H., Misra, I., Jegou, H., Mairal, J., Bojanowski, P., & Joulin, A. (2021). Emerging Properties in Self-Supervised Vision Transformers. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 9650–9660.
别名	SSL semantic segmentation, unsupervised semantic segmentation, label-free semantic segmentation, self-supervised dense prediction	SSL-ViT, self-supervised ViT, unsupervised ViT pre-training, vision transformer self-supervised pre-training
相关	5	4
摘要	Self-supervised semantic segmentation learns to assign a class label to every pixel of an image without relying on manually annotated segmentation masks. A backbone network is first trained on large quantities of unlabeled images using self-supervised objectives such as contrastive learning or masked image modeling, and the resulting dense features are then used to partition and label image regions, achieving competitive segmentation quality at a fraction of the annotation cost.	Self-supervised Vision Transformer (SSL-ViT) applies self-supervised pre-training objectives — such as masked patch prediction (MAE) or self-distillation with no labels (DINO) — to the Vision Transformer architecture, enabling powerful visual representations to be learned from large unlabeled image corpora before any task-specific fine-tuning.