Native Segmentation Vision Transformers
This work addresses the need for more efficient and effective segmentation in computer vision, offering a novel paradigm that could impact various applications, though it appears incremental in advancing transformer-based methods.
The paper tackles the problem of uniform downsampling in vision backbones by proposing a content-aware spatial grouping layer that dynamically reduces tokens based on image boundaries and semantics, resulting in hierarchical segmentation without additional heads. This approach enables strong zero-shot segmentation without mask supervision and efficient model design for downstream tasks.
Uniform downsampling remains the de facto standard for reducing spatial resolution in vision backbones. In this work, we propose an alternative design built around a content-aware spatial grouping layer, that dynamically assigns tokens to a reduced set based on image boundaries and their semantic content. Stacking our grouping layer across consecutive backbone stages results in hierarchical segmentation that arises natively in the feature extraction process, resulting in our coined Native Segmentation Vision Transformer. We show that a careful design of our architecture enables the emergence of strong segmentation masks solely from grouping layers, that is, without additional segmentation-specific heads. This sets the foundation for a new paradigm of native, backbone-level segmentation, which enables strong zero-shot results without mask supervision, as well as a minimal and efficient standalone model design for downstream segmentation tasks. Our project page is https://research.nvidia.com/labs/dvl/projects/native-segmentation.