Journal of Geo-information Science

MDSNet : A Multi-Scale Depth Supervision Method for High-Resolution Remote Sensing Image Semantic Segmentation

SHAN Huilin, WANG Xingtao, LIU Wenxing, WU Xinyue, GAO Runze, LI Hongxu

Journal of Geo-information Science, 2025, 27(6): 1381-1400. DOI: 10.12082/dqxxkx.2025.250009

模型	不透表面		建筑物		低矮植被		树木		汽车		mIoU	mF1	OA
模型	IoU	F1	IoU	F1	IoU	F1	IoU	F1	IoU	F1	mIoU	mF1	OA
CMTFNet	82.46	90.39	60.53	75.41	64.59	78.49	82.33	90.31	76.33	86.58	75.37	85.61	89.97
DeepLabV3+	81.92	90.06	84.42	91.55	62.84	77.18	80.8	89.38	73.43	84.68	77.94	87.38	89.12
MAResU-Net	81.14	89.59	82.07	90.15	68.01	80.96	81.69	89.92	74.25	85.22	78.73	87.98	89.40
EIGNet	81.65	89.90	86.82	92.95	69.16	81.77	80.30	89.07	72.66	84.17	79.22	88.27	89.06
A2FPN	82.80	90.59	87.80	93.50	66.93	80.19	81.90	90.05	74.32	85.27	79.74	88.55	89.66
CMLFormer	82.62	90.48	87.17	93.15	69.03	81.68	81.42	89.76	74.52	85.40	79.94	88.71	89.64
U-Net	83.25	90.86	88.64	84.62	73.34	84.42	82.65	90.50	76.16	86.47	81.62	89.79	90.29
PSPNet	84.36	91.52	87.57	93.37	73.07	84.44	84.64	91.68	78.64	88.22	82.30	90.21	91.01
CGGLNet	84.35	91.51	91.75	95.70	74.72	85.53	84.55	91.63	79.09	88.33	83.49	90.91	91.25
MDSNet	87.16	93.14	94.54	97.19	78.16	87.74	88.74	94.03	83.08	90.76	86.49	92.68	92.87

Tab. 4 Segmentation results under different network models on the Vaihingen dataset (%)

Other figure/table from this article