SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers

TL;DR

SegFormer introduces a simple, efficient transformer-based framework for semantic segmentation that unifies hierarchical encoding with lightweight decoders, achieving state-of-the-art results with fewer parameters.

Contribution

It proposes a novel hierarchical Transformer encoder without positional encoding and a lightweight MLP decoder, improving efficiency and performance over previous methods.

Findings

SegFormer-B4 achieves 50.3% mIoU on ADE20K with 64M parameters.

SegFormer-B5 reaches 84.0% mIoU on Cityscapes.

The model is robust in zero-shot scenarios on Cityscapes-C.

Abstract

We present SegFormer, a simple, efficient yet powerful semantic segmentation framework which unifies Transformers with lightweight multilayer perception (MLP) decoders. SegFormer has two appealing features: 1) SegFormer comprises a novel hierarchically structured Transformer encoder which outputs multiscale features. It does not need positional encoding, thereby avoiding the interpolation of positional codes which leads to decreased performance when the testing resolution differs from training. 2) SegFormer avoids complex decoders. The proposed MLP decoder aggregates information from different layers, and thus combining both local attention and global attention to render powerful representations. We show that this simple and lightweight design is the key to efficient segmentation on Transformers. We scale our approach up to obtain a series of models from SegFormer-B0 to SegFormer-B5, reaching significantly better performance and efficiency than previous counterparts. For example, SegFormer-B4 achieves 50.3% mIoU on ADE20K with 64M parameters, being 5x smaller and 2.2% better than the previous best method. Our best model, SegFormer-B5, achieves 84.0% mIoU on Cityscapes validation set and shows excellent zero-shot robustness on Cityscapes-C. Code will be released at: github.com/NVlabs/SegFormer.