DCP-CLIP:A Coarse-to-Fine Framework for Open-Vocabulary Semantic Segmentation with Dual Interaction

Benchmark Model Rank Results
open-vocabulary-semantic-segmentation-on-ade20k-150DCP-CLIP (CLIP ViT-B/16)–mIoU: 32.4
open-vocabulary-semantic-segmentation-on-ade20k-150DCP-CLIP (CLIP ViT-L/14)–mIoU: 38.1
open-vocabulary-semantic-segmentation-on-ade20k-847DCP-CLIP (CLIP ViT-B/16)–mIoU: 12.3
open-vocabulary-semantic-segmentation-on-ade20k-847DCP-CLIP (CLIP ViT-L/14)–mIoU: 16.4
open-vocabulary-semantic-segmentation-on-pascal-context-459DCP-CLIP (CLIP ViT-B/16)–mIoU: 19.7
open-vocabulary-semantic-segmentation-on-pascal-context-459DCP-CLIP (CLIP ViT-L/14)–mIoU: 24.3
open-vocabulary-semantic-segmentation-on-pascal-context-59DCP-CLIP (CLIP ViT-B/16)–mIoU: 58.1
open-vocabulary-semantic-segmentation-on-pascal-context-59DCP-CLIP (CLIP ViT-L/14)–mIoU: 63.8
open-vocabulary-semantic-segmentation-on-pascalvoc-20DCP-CLIP (CLIP ViT-B/16)–mIoU: 95.9
open-vocabulary-semantic-segmentation-on-pascalvoc-20DCP-CLIP (CLIP ViT-L/14)–mIoU: 97.3