Efficient Transformer Encoder Design for Universal Image Segmentation

A novel transformer encoder design called PRO-SCALE that progressively scales the token length across encoder layers to significantly reduce computational cost while maintaining competitive performance on universal segmentation tasks.