Panoptic Segmentation

Bestes Modell: MasQCLIP

Bestes Modell: OneFormer (DiNAT-L, single-scale, 1280x1280, COCO-Pretrain)

Bestes Modell: OneFormer (ConvNeXt-L, single-scale, Mapillary Vistas-Pretrained)

Bestes Modell: Panoptic FCN* (Swin-L, Cityscapes-fine)

Bestes Modell: OpenSeeD (SwinL, single-scale)

Bestes Modell: VAN-B6*

Bestes Modell: Mask DINO (single scale)

Bestes Modell: SuperCluster

Bestes Modell: EfficientPS

Bestes Modell: EfficientPS

Bestes Modell: Mask2Former (Swin-B)

Bestes Modell: OneFormer (DiNAT-L, single-scale)

Bestes Modell: LKCell

Bestes Modell: (AF)2-S3Net + CenterPoint

Bestes Modell: Exchanger+Mask2Former

Bestes Modell: Early Fusion

Bestes Modell: OneFormer3D

Bestes Modell: OneFormer3D

Bestes Modell: P3Former