Using AAEHS-Net as an Attention-Based Auxiliary Extraction and Hybrid Subsampled Network for Semantic Segmentation

Figure 19

The segmentation results of four currently popular models for two sample images on the Massachusetts roads dataset. (a) Image, (b) mask, (c) MHSU-Net, (d) HAD-ResUNet, (e) MR-UNet, (f) CE-UNet, and (g) AAEHS-Net.

(a)

(b)

(c)

(d)

(e)

(f)

(g)