Vision Transformer Adapter for Dense Predictions
This work addresses the problem of adapting plain ViTs for dense predictions in computer vision, offering a competitive alternative to vision-specific transformers, though it is incremental as it builds on existing ViT frameworks.
The paper tackles the inferior performance of plain Vision Transformers (ViT) on dense prediction tasks by proposing ViT-Adapter, which introduces image-related inductive biases without pre-training, achieving state-of-the-art results such as 60.9 box AP and 53.0 mask AP on COCO test-dev.
This work investigates a simple yet powerful dense prediction task adapter for Vision Transformer (ViT). Unlike recently advanced variants that incorporate vision-specific inductive biases into their architectures, the plain ViT suffers inferior performance on dense predictions due to weak prior assumptions. To address this issue, we propose the ViT-Adapter, which allows plain ViT to achieve comparable performance to vision-specific transformers. Specifically, the backbone in our framework is a plain ViT that can learn powerful representations from large-scale multi-modal data. When transferring to downstream tasks, a pre-training-free adapter is used to introduce the image-related inductive biases into the model, making it suitable for these tasks. We verify ViT-Adapter on multiple dense prediction tasks, including object detection, instance segmentation, and semantic segmentation. Notably, without using extra detection data, our ViT-Adapter-L yields state-of-the-art 60.9 box AP and 53.0 mask AP on COCO test-dev. We hope that the ViT-Adapter could serve as an alternative for vision-specific transformers and facilitate future research. The code and models will be released at https://github.com/czczup/ViT-Adapter.