SeMask: Semantically Masked Transformers for Semantic Segmentation

TL;DR

SeMask introduces a semantic attention framework that enhances hierarchical transformer encoders for semantic segmentation by incorporating semantic priors, leading to state-of-the-art results on ADE20K and Cityscapes datasets.

Contribution

The paper proposes SeMask, a novel semantic attention method that integrates semantic priors into transformer encoders during finetuning, improving segmentation performance.

Findings

Achieved 58.25% mIoU on ADE20K, setting a new state-of-the-art.

Improved Cityscapes mIoU by over 3%.

Enhanced encoder performance with minimal increase in FLOPs.

Abstract

Finetuning a pretrained backbone in the encoder part of an image transformer network has been the traditional approach for the semantic segmentation task. However, such an approach leaves out the semantic context that an image provides during the encoding stage. This paper argues that incorporating semantic information of the image into pretrained hierarchical transformer-based backbones while finetuning improves the performance considerably. To achieve this, we propose SeMask, a simple and effective framework that incorporates semantic information into the encoder with the help of a semantic attention operation. In addition, we use a lightweight semantic decoder during training to provide supervision to the intermediate semantic prior maps at every stage. Our experiments demonstrate that incorporating semantic priors enhances the performance of the established hierarchical encoders with a slight increase in the number of FLOPs. We provide empirical proof by integrating SeMask into Swin Transformer and Mix Transformer backbones as our encoder paired with different decoders. Our framework achieves a new state-of-the-art of 58.25% mIoU on the ADE20K dataset and improvements of over 3% in the mIoU metric on the Cityscapes dataset. The code and checkpoints are publicly available at https://github.com/Picsart-AI-Research/SeMask-Segmentation .