Panoptic Segmentation

Best model: MasQCLIP

Best model: OneFormer (DiNAT-L, single-scale, 1280x1280, COCO-Pretrain)

Best model: OneFormer (ConvNeXt-L, single-scale, Mapillary Vistas-Pretrained)

Best model: Panoptic FCN* (Swin-L, Cityscapes-fine)

Best model: OpenSeeD (SwinL, single-scale)

Best model: VAN-B6*

Best model: Mask DINO (single scale)

Best model: SuperCluster

Best model: EfficientPS

Best model: EfficientPS

Best model: Mask2Former (Swin-B)

Best model: OneFormer (DiNAT-L, single-scale)

Best model: LKCell

Best model: (AF)2-S3Net + CenterPoint

Best model: Exchanger+Mask2Former

Best model: Early Fusion

Best model: OneFormer3D

Best model: OneFormer3D

Best model: P3Former