Table 8 Comparison of no-reference image quality metrics for various enhancement techniques on the VV dataset84.
From: A multi scale spatial attention based zero shot learning framework for low light image enhancement
Metric | Paired supervision | Unsupervised | Zero-shot learning | ||||||
|---|---|---|---|---|---|---|---|---|---|
BIMEF102 | LIME3 | MF103 | Multiscale Retinex104 | EnlightenGAN56 | SG-ZSL54 | Zero-DCE52 | Zero-DCE++53 | Ours | |
NIMA91 | 4.62 | 4.62 | 4.69 | 4.69 | 4.32 | 4.32 | 4.18 | 4.19 | 4.73 |
PaQ2PiQ92 | 74.16 | 76.41 | 74.84 | 75.60 | 72.69 | 72.66 | 73.70 | 73.71 | 76.52 |
DBCNN93 | 58.44 | 55.18 | 59.35 | 55.54 | 56.03 | 56.16 | 46.21 | 46.22 | 59.68 |
musiq-koniq55 | 65.40 | 64.49 | 65.98 | 63.57 | 61.08 | 61.11 | 58.78 | 58.82 | 66.10 |
MANIQA94 | 0.66 | 0.65 | 0.67 | 0.60 | 0.64 | 0.64 | 0.62 | 0.62 | 0.68 |
CLIPIQA95 | 0.50 | 0.51 | 0.52 | 0.46 | 0.54 | 0.55 | 0.47 | 0.47 | 0.57 |
TReS-koniq96 | 77.48 | 68.16 | 77.63 | 75.08 | 74.63 | 74.35 | 66.64 | 66.68 | 78.10 |
HyperIQA97 | 0.61 | 0.58 | 0.61 | 0.59 | 0.51 | 0.51 | 0.43 | 0.44 | 0.63 |
GPR-BIQA10 | 0.66 | 0.63 | 0.67 | 0.62 | 0.60 | 0.61 | 0.58 | 0.59 | 0.69 |
Quality Net98 | 0.67 | 0.66 | 0.68 | 0.66 | 0.64 | 0.65 | 0.61 | 0.62 | 0.70 |
PIQI9 | 0.66 | 0.65 | 0.67 | 0.65 | 0.63 | 0.63 | 0.60 | 0.61 | 0.68 |
Average | 25.81 | 24.78 | 26.03 | 25.28 | 24.76 | 24.74 | 22.98 | 23.01 | 26.28 |
