Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers

Code

• github.com/fudan-zvg/SETR
  Officialpytorch★ 1,109
• github.com/PaddlePaddle/PaddleSeg
  paddle★ 9,319
• github.com/gupta-abhay/setr-pytorch
  pytorch★ 201
• github.com/920232796/setr-pytorch
  pytorch★ 133
• github.com/BR-IDL/PaddleViT/tree/main/semantic_segmentation
  paddle★ 0

Abstract

Most recent semantic segmentation methods adopt a fully-convolutional network (FCN) with an encoder-decoder architecture. The encoder progressively reduces the spatial resolution and learns more abstract/semantic visual concepts with larger receptive fields. Since context modeling is critical for segmentation, the latest efforts have been focused on increasing the receptive field, through either dilated/atrous convolutions or inserting attention modules. However, the encoder-decoder based FCN architecture remains unchanged. In this paper, we aim to provide an alternative perspective by treating semantic segmentation as a sequence-to-sequence prediction task. Specifically, we deploy a pure transformer (ie, without convolution and resolution reduction) to encode an image as a sequence of patches. With the global context modeled in every layer of the transformer, this encoder can be combined with a simple decoder to provide a powerful segmentation model, termed SEgmentation TRansformer (SETR). Extensive experiments show that SETR achieves new state of the art on ADE20K (50.28% mIoU), Pascal Context (55.83% mIoU) and competitive results on Cityscapes. Particularly, we achieve the first position in the highly competitive ADE20K test server leaderboard on the day of submission.

Tasks

Benchmark Results

Reproductions