Generalized Decoding for Pixel, Image, and Language

Open paper
Benchmark Model Rank Results
instance-segmentation-on-ade20k-valX-Decoder (Davit-d5, Deform, single-scale, 1280x1280)#5AP: 38.7APL: 59.6APM: 43.3APS: 18.9
instance-segmentation-on-ade20k-valX-Decoder (L)#9AP: 35.8
panoptic-segmentation-on-ade20k-valX-Decoder (Davit-d5, Deform, single-scale, 1280x1280)#6PQ: 52.4mIoU: 59.1AP: 38.7
panoptic-segmentation-on-ade20k-valX-Decoder (L)#15PQ: 49.6mIoU: 58.1AP: 35.8
referring-expression-segmentation-on-refcocog-valX-Decoder (Davit-d5)#15Overall IoU: 64.6
zero-shot-segmentation-on-segmentation-in-the-wildSGinW_Team (X-Decoder-L)#8Mean AP: 32.2
zero-shot-segmentation-on-segmentation-in-the-wildSGinW_Team (X-Decoder-B)#9Mean AP: 27.7
zero-shot-segmentation-on-segmentation-in-the-wildSGinW_Team (X-Decoder-L-IN21K)#10Mean AP: 26.6
zero-shot-segmentation-on-segmentation-in-the-wildSGinW_Team (X-Decoder-T)#11Mean AP: 22.6

Rank counts only results with a code link.