Learning Mask-aware CLIP Representations for Zero-Shot Segmentation

Benchmark Model Rank Results
open-vocabulary-semantic-segmentation-on-pascalvoc-20MAFT-ViTL#11mIoU: 92.1
open-vocabulary-semantic-segmentation-on-pascal-context-59MAFT-ViTL#13mIoU: 58.5
open-vocabulary-semantic-segmentation-on-pascal-context-459MAFT-ViTL#14mIoU: 15.7
open-vocabulary-semantic-segmentation-on-ade20k-150MAFT-ViTL#16mIoU: 32.0
open-vocabulary-semantic-segmentation-on-ade20k-847MAFT-ViTL#17mIoU: 12.1