Panoptic Segmentation On Cityscapes Val

评估指标

PQst

PQth

mIoU

评测结果

各个模型在此基准测试上的表现结果

						Paper Title	Repository
OneFormer (ConvNeXt-L, single-scale, 512x1024, Mapillary Vistas-pretrained)	48.7	70.1	74.1	64.6	84.6	OneFormer: One Transformer to Rule Universal Image Segmentation
Panoptic-DeepLab (SWideRNet [1, 1, 4.5], Mapillary Vistas, multi-scale)	46.8	69.6	-	-	85.3	Scaling Wide Residual Networks for Panoptic Segmentation	-
OneFormer (ConvNeXt-L, single-scale)	46.5	68.51	-	-	83.0	OneFormer: One Transformer to Rule Universal Image Segmentation
Axial-DeepLab-XL (Mapillary Vistas, multi-scale)	44.2	68.5	-	-	84.6	Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation
Panoptic-DeepLab (SWideRNet [1, 1, 4.5], Mapillary Vistas, single-scale)	42.8	68.5	-	-	84.6	Scaling Wide Residual Networks for Panoptic Segmentation	-
OneFormer (ConvNeXt-XL, single-scale)	46.7	68.4	-	-	83.6	OneFormer: One Transformer to Rule Universal Image Segmentation
kMaX-DeepLab (single-scale)	44.0	68.4	-	-	83.5	kMaX-DeepLab: k-means Mask Transformer
AFF-Base (single-scale, point-based Mask2Former)	46.2	67.7	71.5	62.5	83.0	AutoFocusFormer: Image Segmentation off the Grid
OneFormer (DiNAT-L, single-scale)	45.6	67.6	-	-	83.1	OneFormer: One Transformer to Rule Universal Image Segmentation
EfficientPS	43.5	67.5	70.3	63.2	82.1	EfficientPS: Efficient Panoptic Segmentation
DiNAT-L (Mask2Former)	44.5	67.2	-	-	83.4	Dilated Neighborhood Attention Transformer
OneFormer (Swin-L, single-scale)	45.6	67.2	-	-	83.0	OneFormer: One Transformer to Rule Universal Image Segmentation
AFF-Small (single-scale, point-based Mask2Former)	44.2	66.9	70.8	61.5	82.2	AutoFocusFormer: Image Segmentation off the Grid
Mask2Former (Swin-L)	43.6	66.6	-	-	82.9	Masked-attention Mask Transformer for Universal Image Segmentation
EfficientPS (Cityscapes-fine)	39.1	64.9	67.7	61.0	90.3	EfficientPS: Efficient Panoptic Segmentation
CMT-DeepLab (MaX-S, single-scale, IN-1K)	-	64.6	-	-	81.4	CMT-DeepLab: Clustering Mask Transformers for Panoptic Segmentation
Panoptic-DeepLab (X71)	38.5	64.1	-	-	81.5	Panoptic-DeepLab: A Simple, Strong, and Fast Baseline for Bottom-Up Panoptic Segmentation
Mask2Former + Intra-Batch Supervision (ResNet-50)	-	62.4	67.3	54.7	-	Intra-Batch Supervision for Panoptic Segmentation on High-Resolution Images
COPS (ResNet-50)	34.1	62.1	67.2	55.1	79.3	Combinatorial Optimization for Panoptic Segmentation: A Fully Differentiable Approach
AdaptIS (ResNeXt-101)	36.3	62.0	64.4	58.7	79.2	AdaptIS: Adaptive Instance Selection Network	-

0 of 36 row(s) selected.

Command Palette

Panoptic Segmentation On Cityscapes Val

评估指标

评测结果