Efficient Self-supervised Vision Transformers for Representation Learning
This work addresses efficiency and representation quality in self-supervised learning for computer vision, offering a novel method that improves performance over prior arts, though it is incremental in combining existing techniques.
The paper tackles the problem of reducing complexity in self-supervised vision transformers while maintaining fine-grained region correspondence, achieving 81.3% top-1 accuracy on ImageNet linear probe with significantly higher throughput and outperforming supervised models on most downstream datasets.
This paper investigates two techniques for developing efficient self-supervised vision transformers (EsViT) for visual representation learning. First, we show through a comprehensive empirical study that multi-stage architectures with sparse self-attentions can significantly reduce modeling complexity but with a cost of losing the ability to capture fine-grained correspondences between image regions. Second, we propose a new pre-training task of region matching which allows the model to capture fine-grained region dependencies and as a result significantly improves the quality of the learned vision representations. Our results show that combining the two techniques, EsViT achieves 81.3% top-1 on the ImageNet linear probe evaluation, outperforming prior arts with around an order magnitude of higher throughput. When transferring to downstream linear classification tasks, EsViT outperforms its supervised counterpart on 17 out of 18 datasets. The code and models are publicly available: https://github.com/microsoft/esvit