Semantic Segmentation On Isaid

Metrics

mIoU

Results

Performance results of various models on this benchmark

		Paper Title
SegNeXt-L	70.3	SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation
SegNeXt-B	69.9	SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation
AerialFormer-B	69.3	AerialFormer: Multi-resolution Transformer for Aerial Image Segmentation
SegNeXt-S	68.8	SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation
AerialFormer-S	68.4	AerialFormer: Multi-resolution Transformer for Aerial Image Segmentation
SegNeXt-T	68.3	SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation
FarSeg++@MiT-B2	67.9	FarSeg++: Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
FarSeg++@ResNet-50	67.6	FarSeg++: Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
AerialFormer-T	67.5	AerialFormer: Multi-resolution Transformer for Aerial Image Segmentation
DeepLabV3 with R-50	67.03	Resolution-Aware Design of Atrous Rates for Semantic Segmentation Networks
FarSeg++@Swin-T	66.3	FarSeg++: Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
IMP-ViTAEv2-S-UperNet	65.3	An Empirical Study of Remote Sensing Pretraining
FactSeg@ResNet-50	64.79	FactSeg: Foreground Activation Driven Small Object Semantic Segmentation in Large-Scale Remote Sensing Imagery
ViTAE-B + RVSA-UperNet	64.49	Advancing Plain Vision Transformer Towards Remote Sensing Foundation Model
RSP-ViTAEv2-S-UperNet	64.3	An Empirical Study of Remote Sensing Pretraining
RSP-Swin-T-UperNet	64.1	An Empirical Study of Remote Sensing Pretraining
ViT-B + RVSA-UperNet	63.85	Advancing Plain Vision Transformer Towards Remote Sensing Foundation Model
FarSeg@ResNet-50	63.71	Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
RSP-ResNet-50-UperNet	61.6	An Empirical Study of Remote Sensing Pretraining

0 of 19 row(s) selected.