River Segmentation of Remote Sensing Images Based on Composite Attention Network

Table 4

Comparison of evaluation indexes of different river segmentation networks.

Network	PA	MPA	MIoU	FMIoU
FCN-8s	0.9524	0.9236	0.9086	0.9462
SegNet	0.9602	0.9311	0.9172	0.9548
BiSeNet	0.9617	0.9335	0.9234	0.9603
U-Net	0.9682	0.9472	0.9418	0.9674
D-LinkNet	0.9691	0.9467	0.9407	0.9641
CoANet	0.9743	0.9556	0.9502	0.9728

Bold indicates the maximum value of each evaluation index.