OTSeg: Multi-prompt Sinkhorn Attention for Zero-Shot Semantic Segmentation

Benchmark Model Rank Results
zero-shot-semantic-segmentation-on-coco-stuffOTSeg+#1Transductive Setting hIoU: 49.8Inductive Setting hIoU: 41.5
zero-shot-semantic-segmentation-on-coco-stuffOTSeg#3Transductive Setting hIoU: 49.5Inductive Setting hIoU: 41.4
zero-shot-semantic-segmentation-on-pascal-vocOTSeg+#1Transductive Setting hIoU: 94.4Inductive Setting hIoU: 87.4
zero-shot-semantic-segmentation-on-pascal-vocOTSeg#2Transductive Setting hIoU: 94.2Inductive Setting hIoU: 84.5