Panoptic Segmentation On Mapillary Val

Metrics

PQ

PQst

PQth

mIoU

Results

Performance results of various models on this benchmark

					Paper Title
OneFormer (DiNAT-L, single-scale)	46.7	54.9	40.5	61.7	OneFormer: One Transformer to Rule Universal Image Segmentation
OneFormer (ConvNeXt-L, single-scale)	46.4	54.0	40.6	61.6	OneFormer: One Transformer to Rule Universal Image Segmentation
Panoptic FCN* (Swin-L, single-scale)	45.7	52.1	40.8	-	Fully Convolutional Networks for Panoptic Segmentation
Panoptic-DeepLab (SWideRNet-(1, 1, 4.5), multi-scale)	44.8	51.9	39.3	60.0	Scaling Wide Residual Networks for Panoptic Segmentation
Mask2Former + Intra-Batch Supervision (ResNet-50)	42.2	52.0	34.9	-	Intra-Batch Supervision for Panoptic Segmentation on High-Resolution Images
Axial-DeepLab-L (multi-scale)	41.1	51.3	33.4	58.4	Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation
EfficientPS	40.6	-	-	-	EfficientPS: Efficient Panoptic Segmentation
Panoptic-DeepLab (X71)	40.5	-	-	-	Panoptic-DeepLab: A Simple, Strong, and Fast Baseline for Bottom-Up Panoptic Segmentation
AdaptIS (ResNeXt-101)	40.3	-	-	56.8	AdaptIS: Adaptive Instance Selection Network
Panoptic FCN* (ResNet-FPN)	36.9	-	32.9	-	Fully Convolutional Networks for Panoptic Segmentation
JSIS-Net (ResNet-50)	17.6	-	-	-	Panoptic Segmentation with a Joint Semantic and Instance Segmentation Network
HRNet-OCR (Hierarchical Multi-Scale Attention)	17.6	-	-	-	Hierarchical Multi-Scale Attention for Semantic Segmentation
Panoptic FCN* (ResNet-50-FPN)	-	42.3	-	-	Fully Convolutional Networks for Panoptic Segmentation

0 of 13 row(s) selected.

Panoptic Segmentation On Mapillary Val | SOTA | HyperAI