Module segformer

Module segformer 

Source
Expand description

Segformer model implementation for semantic segmentation and image classification.

Segformer is a transformer-based model designed for vision tasks. It uses a hierarchical structure that progressively generates features at different scales.

Key characteristics:

  • Efficient self-attention with sequence reduction
  • Hierarchical feature generation
  • Mix-FFN for local and global feature interaction
  • Lightweight all-MLP decode head

References:

Structsยง

Config
ImageClassificationModel
SemanticSegmentationModel