Table 11 Comparison of semantic segmentation performance after processing with low-light enhancement algorithms (1500 images).
From: Content-illumination coupling guided low-light image enhancement network
Methods\index | DeepLabv3 + _mobilenet | DeepLabv3 + _resnet101 | ||
|---|---|---|---|---|
mPA↑/(%) | mIoU↑/(%) | mPA↑/(%) | mIoU↑/(%) | |
Segmentation of low-light samples | 29.10 | 23.00 | 34.40 | 26.85 |
HDRNet | 21.55 | 16.76 | 27.23 | 19.62 |
SARN | 33.37 | 26.90 | 38.59 | 29.84 |
ALEN | 27.51 | 21.71 | 32.23 | 25.39 |
ABSGNet | 34.85 | 25.69 | 37.03 | 28.43 |
MPRNet | 34.66 | 25.31 | 37.46 | 28.77 |
Restormer | 34.91 | 27.35 | 37.72 | 30.31 |
CICGNet | 36.68 | 30.51 | 38.76 | 32.79 |
