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

Segmentation Transformer

Implementation of Segmentation Transformer in PyTorch, a new model to achieve SOTA in semantic segmentation while using transformer style encoders.

Features

• [x] SETR
  • [x] SETR-Naive
  • [x] SETR-PUP
  • [x] SETR-MLA
• [x] SETR-Hybrid

To Do:

• [ ] Training Scripts

Installation

Create the environment:

conda env create -f environment.yml

GitHub