Deep Feature Prior Guided Face Deblurring

Semantic-Aware Face Deblurring With Pixel-Wise Projection Discriminator

“…Single image deblurring has been studied extensively over the past decades. In this section, we focus our discussion on recent deep learning (DL)-based deblurring methods, which can be divided into two categories: general image deblurring [16], [18], [19], [27], (b) Projection GAN [28], (c) U-Net GAN [24], (d) proposed SAPPGAN.…”

“…The associate editor coordinating the review of this manuscript and approving it for publication was Taous Meriem Laleg-Kirati . maps [16], [17], [18], and deep features [19]. Despite these efforts, these methods [13], [15], [17] often suffer from over-smoothed and perceptually unnatural results.…”

“…Despite these efforts, these methods [13], [15], [17] often suffer from over-smoothed and perceptually unnatural results. Some SFID methods [14], [16], [18], [19] are effective at improving the perceptual qualities of deblurred images on the strength of generative adversarial networks (GANs) [20]. GANs have demonstrated an ability to generate realistic samples via a min-max game between a generator and a discriminator.…”

“…1 (a). Several SFID methods [14], [16], [18], [19] leverage this distribution modeling ability to impose a constraint that the deblurred image should follow the distribution of sharp ground-truth (GT) images.…”

All One Needs to Know about Priors for Deep Image Restoration and Enhancement: A Survey

“…The reference-based prior relies on the guidance from the highquality face images [154], [161] or facial component dictionaries [160]. The reference priors are always obtained by applying [154] ECCV 2018 Face Restoration Reference-based Prior High-quality guided image of the same identity Yu [153] ECCV 2018 Face Hallucination (SR) Facial Component Facial component heatmaps Chen [133] CVPR 2018 Face Hallucination (SR) Geometry Facial Landmarks / Parsing Maps Bulat [157] CVPR 2018 Face Hallucination (SR) Geometry End-to-end face SR and landmark localization Chrysos [158] IJCV 2019 Face Deblurring Geometry Facial Landmarks / Parsing Maps Kim [159] BMVC2019 Face Hallucination (SR) Geometry Facial Landmarks Li [160] ECCV 2020 Face Restoration Reference-based Prior Deep face components dictionaries Ma [144] CVPR 2020 Face Hallucination (SR) Geometry Facial Landmarks / Parsing Maps Li [161] CVPR 2020 Face Hallucination (SR) Reference-based Prior High quality images as reference Kalarot [162] WACV 2020 Face Hallucination (SR) Facial Component Facial component-wise attention maps Shen [163] IJCV 2020 Face Deblurring Facial Component Semantic labels priors and local structures prior Yasarla [156] TIP 2020 Face Deblurring Facial Component Semantic labels priors Grm [164] TIP 2020 Face Hallucination (SR) Recognition Model Identity prior from face recognition model Kim [165] ICIP 2021 Face Hallucination (SR) Facial Component Non-parametric facial prior Chen [139] NeurIPS 2021 Face Restoration Parsing Maps Semantic aware style transformation Hu [155] TPAMI 2021 Face Restoration Geometry Plug-and-play 3D facial prior Chen [139] CVPR 2021 Face Restoration Geometry Parsing Maps Wang [23] CVPR 2021 Face Restoration GAN Facial Prior Prior in a pre-trained face GAN Yang [166] CVPR 2021 Face Restoration GAN Facial Prior Fine-tune the GAN prior embedded DNN Jung [167] WACV 2022 Face Deblurring Recognition Model Deep features of face recognition networks traditional methods, e.g., K-means, to cluster the perceptually significant characteristics of facial components. These priors are then applied to tackle more challenging face image restoration problems.…”