Table 5 LOE and DE scores on MEF, LIME, NPE, DICM datasets, respectively.

From: Low-light image enhancement via adaptive frequency decomposition network

Methods

MEF

LIME

NPE

DICM

AVG

LOE\(\downarrow \)

DE\(\uparrow \)

LOE\(\downarrow \)

DE\(\uparrow \)

LOE\(\downarrow \)

DE\(\uparrow \)

LOE\(\downarrow \)

DE\(\uparrow \)

LOE\(\downarrow \)

DE\(\uparrow \)

MBLLEN17

256.54

7.14

214.65

7.1770

267.45

7.1461

396.62

7.32

321.32

7.19

Retinex-Net16

473.85

6.92

508.73

7.08

849.12

6.82

600.11

7.17

579.59

7.00

Kind18

378.41

6.71

302.41

6.91

354.98

6.85

305.33

6.28

334.20

6.69

R2RNet18

270.55

6.52

177.26

7.31

212.51

7.14

356.05

6.81

393.05

6.94

SCI29

389.66

5.53

247.12

6.8741

274.12

6.81

430.83

6.31

374.94

6.38

RUAS28

528.40

5.90

387.23

5.93

342.15

5.78

334.52

6.28

467.44

5.97

EnlightenGAN13

443.76

6.83

183.76

7.29

208.34

7.22

653.94

7.31

572.83

7.16

Zero-DCE12

398.44

6.15

391.00

7.13

652.13

6.86

594.28

7.24

320.38

6.84

Zero-DCE++20

265.48

6.25

184.26

7.24

426.45

7.14

365.44

7.29

335.34

6.98

HFMNet21

303.39

7.13

231.46

7.22

294.04

7.05

364.52

7.14

309.03

7.14

Kind++19

356.21

7.00

198.78

7.15

231.24

7.08

420.87

7.20

377.23

7.11

Ours

235.67

7.23

158.05

7.43

195.41

7.37

358.67

7.46

283.04

7.37

  1. The best results are highlighted in bold.