SegFormer-inference-template

In this repository, I implemented semantic segmentation inference code by using SegFormer model from hugging face. I hope this repository will be helpful when preprocessing to mask a particular area.

model

You can choose a pre-trained segformer model from the following list, and designate when running segmentation.py --model model_name

[
  'nvidia/segformer-b5-finetuned-cityscapes-1024-1024',  
  'nvidia/segformer-b5-finetuned-ade-640-640',
  'nvidia/segformer-b4-finetuned-cityscapes-1024-1024',  
  'nvidia/segformer-b4-finetuned-ade-512-512',
  'nvidia/segformer-b3-finetuned-cityscapes-1024-1024',  
  'nvidia/segformer-b3-finetuned-ade-512-512',
  'nvidia/segformer-b2-finetuned-cityscapes-1024-1024',  
  'nvidia/segformer-b2-finetuned-ade-512-512',
  'nvidia/segformer-b1-finetuned-cityscapes-1024-1024',  
  'nvidia/segformer-b1-finetuned-ade-512-512',
  'nvidia/segformer-b0-finetuned-cityscapes-1024-1024',
  'nvidia/segformer-b0-finetuned-cityscapes-512-1024',  
  'nvidia/segformer-b0-finetuned-cityscapes-640-1280',  
  'nvidia/segformer-b0-finetuned-cityscapes-768-768',  
  'nvidia/segformer-b0-finetuned-ade-512-512'
]

SegFormer

See this document if you want to learn more about the model