BEiT v2: Masked Image Modeling with Vector-Quantized Visual Tokenizers
This work addresses the limitation of existing MIM methods in exploiting high-level semantics for self-supervised representation learning in computer vision, offering incremental improvements over prior approaches.
The paper tackles the problem of masked image modeling (MIM) being hindered by low-level pixel reconstruction by proposing BEiT v2, which uses a semantic-rich visual tokenizer as the reconstruction target to promote MIM to a semantic level, resulting in improved performance such as 85.5% top-1 accuracy on ImageNet-1K fine-tuning and 56.7% mIoU on ADE20K segmentation.
Masked image modeling (MIM) has demonstrated impressive results in self-supervised representation learning by recovering corrupted image patches. However, most existing studies operate on low-level image pixels, which hinders the exploitation of high-level semantics for representation models. In this work, we propose to use a semantic-rich visual tokenizer as the reconstruction target for masked prediction, providing a systematic way to promote MIM from pixel-level to semantic-level. Specifically, we propose vector-quantized knowledge distillation to train the tokenizer, which discretizes a continuous semantic space to compact codes. We then pretrain vision Transformers by predicting the original visual tokens for the masked image patches. Furthermore, we introduce a patch aggregation strategy which associates discrete image patches to enhance global semantic representation. Experiments on image classification and semantic segmentation show that BEiT v2 outperforms all compared MIM methods. On ImageNet-1K (224 size), the base-size BEiT v2 achieves 85.5% top-1 accuracy for fine-tuning and 80.1% top-1 accuracy for linear probing. The large-size BEiT v2 obtains 87.3% top-1 accuracy for ImageNet-1K (224 size) fine-tuning, and 56.7% mIoU on ADE20K for semantic segmentation. The code and pretrained models are available at https://aka.ms/beitv2.