Semantic Segmentation on UAVid (8 Class IoU Metrics) benchmark leaderboard · SOTA2 Research

Semantic Segmentation on UAVid (8 Class IoU Metrics)

95.56Building IoUPreserve-the-Hard-Regenerate-the-Rest

Evaluation Results

Method
Preserve-the-Hard-Regenerate-the-Rest2026.06	95.56	—	88.47	83.06	77.2	64.57	68.19	32.72	63.48
Preserve-the-Hard-Regenerate-the-Rest2026.06	95.54	—	88.61	82.89	77.08	65.8	69.28	33.32	63.99
Instance Augmentation2026.06	95.52	—	88.09	81.35	74.1	60.79	64.84	33.24	62.24
Preserve-the-Hard-Regenerate-the-Rest2026.06	95.42	—	88.09	82.47	75.94	62.47	64.9	31.38	62.58
Simple Background Augmentation2026.06	95.06	—	87.19	81.01	74.25	58.99	63.81	31.61	61.49
Real data only2026.06	94.98	—	87.32	81.18	73.66	57.05	59.58	28.72	60.31
Dynamic Dictionary Learning2025.03	0.897	0.71	0.832	0.821	0.661	0.75	0.59	0.414	0.709
SFA-Net2025.03	0.89	0.702	0.827	0.808	0.646	0.775	0.675	0.307	0.704
UNetFormer2025.03	0.874	0.684	0.815	0.802	0.635	0.736	0.564	0.31	0.678
ABCNet2025.03	0.864	0.674	0.812	0.799	0.631	0.698	0.484	0.139	0.638
SegFormer2025.03	0.863	0.666	0.801	0.796	0.623	0.725	0.525	0.285	0.66
BANet2025.03	0.854	0.667	0.807	0.789	0.621	0.693	0.528	0.21	0.646
DANet2025.03	0.589	0.649	0.779	0.683	0.615	0.596	0.474	0.091	0.606