Journal of Geo-information Science

Figure/Table detail

Remote Sensing Image Segmentation Based on Multi-modal Feature Extraction and Hierarchical Perception

ZHANG Yinsheng, SHAN Mengjiao, CHEN Xin, CHEN Ge, TONG Junyi, JI Ru, SHAN Huilin

Journal of Geo-information Science, 2024, 26(12): 2741-2758. DOI: 10.12082/dqxxkx.2024.240488

网络	OA/%	P/%	R/%	F1/%	FPS/(帧/s)	MIoU/%
ResNet	81.87	83.61	80.14	81.46	97.38	76.91
V3	85.44	86.03	81.89	82.90	106.21	79.39
DBB	86.54	87.21	82.73	84.92	113.11	80.78

Tab. 5 Comparison of backbone network feature extraction (Vaihingen)

Other figure/table from this article

Fig. 1 The model structure of MFEHPNet
Fig. 2 The model of multi-modal feature extraction
Fig. 3 The mechanism of Coordinate Attention
Fig. 4 The structure of Dual-path Bottleneck Block
Tab. 1 The network structure of MobileNetV3
Fig. 5 The model of Hierarchical Perceptual Network
Fig. 6 The model structure of attention
Fig. 7 The structure of dual fusion
Fig. 8 The model of multi-scale-feature aggregation
Tab. 2 Experimental configuration information
Fig. 9 Examples of ISPRS Vaihingen dataset
Fig. 10 Examples of ISPRS Potsdam dataset
Fig. 11 Difficulties of remote sensing image segmentation
Tab. 3 Experimental comparison of multi-modal fusion methods (Vaihingen) (%)
Tab. 4 Experimental comparison of input different modes (Vaihingen) (%)
Fig. 12 Comparison of input DSM effects
Tab. 6 Comparison of abstract deep feature extraction (Vaihingen) (%)
Tab. 7 Experimental comparison of decoding method extraction (Vaihingen) (%)
Tab. 8 Ablation experiments of each module (Vaihingen)
Fig. 13 Visualization results of ablation experiments
Tab. 9 Comparison of model calculation time
Tab. 10 Comparative experiments of different models (Vaihingen)
Tab. 11 Comparative experiments of different models (Potsdam)
Fig. 14 Segmentation results of low vegetation and trees in Vaihingen
Fig. 15 Segmentation results of buildings and roads in Vaihingen
Fig. 16 Segmentation results of cars in Vaihinge
Fig. 17 Segmentation results in Potsdam

模态框（Modal）标题

Figure/Table detail