Harnessing Vision Foundation Models for High-Performance, Training-Free Open Vocabulary Segmentation

Benchmark Model Rank Results
unsupervised-semantic-segmentation-with-10Trident#2mIoU: 42.2
unsupervised-semantic-segmentation-with-11Trident#3mIoU: 70.8
unsupervised-semantic-segmentation-with-12Trident#3mIoU: 40.1
unsupervised-semantic-segmentation-with-3Trident#2mIoU: 47.6
unsupervised-semantic-segmentation-with-4Trident#3Mean IoU (val): 26.7
unsupervised-semantic-segmentation-with-7Trident#3mIoU: 88.7
unsupervised-semantic-segmentation-with-8Trident#3mIoU: 44.3
unsupervised-semantic-segmentation-with-9Trident#3mIoU: 28.6