segformer¶


Model type	`segformer`
Class	`SegformerForSemanticSegmentation`
Task	`image-classification`
Source	`models/segformer.py`

Description¶

Segformer for semantic segmentation.

Hierarchical ViT encoder with lightweight MLP decoder. Outputs segmentation logits of shape [batch, num_labels, H/4, W/4].

mobius build --model <MODEL_ID> output_dir/

from mobius import build

model = build("<MODEL_ID>")