Face Inpainting Research Articles

Selection and guidance: high-dimensional identity consistency preservation for face inpainting

Selection and guidance: high-dimensional identity consistency preservation for face inpainting

Diverse Semantic Image Synthesis with various conditioning modalities

Semantic image synthesis aims to generate high-fidelity images from a segmentation mask, and previous methods typically train a generator to associate a global random map with the conditioning mask. However, the lack of independent control of regional content impedes their application. To address this issue, we propose an effective approach for Multi-modal conditioning-based Diverse Semantic Image Synthesis, which is referred to as McDSIS. In this model, there are a number of constituent generators incorporated to synthesize the content in semantic regions from independent random maps. The regional content can be determined by the style code associated with a random map, extracted from a reference image, or by embedding a textual description via our proposed conditioning mechanisms. As a result, the generation process is spatially disentangled, which facilitates independent synthesis of diverse content in a semantic region, while at the same time preserving other content. Due to this flexible architecture, in addition to achieving superior performance over state-of-the-art semantic image generation models, McDSIS is capable of performing various visual tasks, such as face inpainting, swapping, local editing, etc.

Driving mutual advancement of 3D reconstruction and inpainting for masked faces

Target occlusion or pollution has always been a common and difficult problem in 3D reconstruction, seriously affecting the reconstruction effect, especially in single image scenario. To address the issues of incomplete reconstruction caused by pixel missing, a novel framework promoting mutual enhancement between 3D reconstruction and inpainting is proposed, which is capable of reconstructing a realistic 3D face with pixel completion from single image without prior assumptions. The framework is composed of an inpainting module for facial pixel completion and a reconstruction module for 3D face modeling. The inpainting module provides complete texture and accurate 3D geometry inferring premise for 3D reconstruction, and the reconstruction module is also involved in face inpainting, constraining the inpainting module to fill plausible pixels that are closer to the ground truth. Experiments show that the accurate 3D faces with complete and fine texture can be reconstructed by the proposed framework from largely masked images, with a competitive performance even surpassing most of state-of-the-art methods with unmasked images as input. Furthermore, this framework has achieved state-of-the-art performance in image inpainting on multiple face datasets.

SwapInpaint2: Towards high structural consistency in identity-guided inpainting via background-preserving GAN inversion

In this work, we propose SwapInpaint2 to enhance the naturalness of identity-guided face inpainting. The previous version, SwapInpaint, relied on a generic inpainting model for content infer, bringing the problems of unnatural structure to the results. Our method addresses this issue by utilizing an inversion-based Background-preserving Attribute Extractor and an improved Embedding Integration Generator to provide high-quality attribute embeddings and produce high-naturalistic results. Comparison experiments with state-of-the-art models demonstrate that our new method achieves superior structural consistency and stronger style alignment both in terms of quality and quantity.

Rdfinet: reference-guided directional diverse face inpainting network

The majority of existing face inpainting methods primarily focus on generating a single result that visually resembles the original image. The generation of diverse and plausible results has emerged as a new branch in image restoration, often referred to as “Pluralistic Image Completion”. However, most diversity methods simply use random latent vectors to generate multiple results, leading to uncontrollable outcomes. To overcome these limitations, we introduce a novel architecture known as the Reference-Guided Directional Diverse Face Inpainting Network. In this paper, instead of using a background image as reference, which is typically used in image restoration, we have used a face image, which can have many different characteristics from the original image, including but not limited to gender and age, to serve as a reference face style. Our network firstly infers the semantic information of the masked face, i.e., the face parsing map, based on the partial image and its mask, which subsequently guides and constrains directional diverse generator network. The network will learn the distribution of face images from different domains in a low-dimensional manifold space. To validate our method, we conducted extensive experiments on the CelebAMask-HQ dataset. Our method not only produces high-quality oriented diverse results but also complements the images with the style of the reference face image. Additionally, our diverse results maintain correct facial feature distribution and sizes, rather than being random. Our network has achieved SOTA results in face diverse inpainting when writing. Code will is available at https://github.com/nothingwithyou/RDFINet.

Unsupervised masked face inpainting based on contrastive learning and attention mechanism

Unsupervised masked face inpainting based on contrastive learning and attention mechanism

SFI-Swin: symmetric face inpainting with swin transformer by distinctly learning face components distributions

SFI-Swin: symmetric face inpainting with swin transformer by distinctly learning face components distributions

ABANet: Attention boundary‐aware network for image segmentation

AbstractDeep learning techniques have attained substantial progress in various face‐related tasks, such as face recognition, face inpainting, and facial expression recognition. To prevent infection or the spread of the virus, wearing of masks in public places has been mandated following the COVID‐19 epidemic, which has led to face occlusion and posed significant challenges for face recognition systems. Most prominent masked face recognition solutions rely on mask segmentation tasks. Therefore, segmentation can be used to mitigate the negative impacts of wearing a mask and improve recognition accuracy. Mask region segmentation suffers from two main problems: there is no standard type of masks that people wear, they come in different colours and designs, and there is no publicly available masked face dataset with appropriate ground truth for the mask region. In order to address these issues, we propose an encoder–decoder framework that utilizes a boundary‐aware attention network combined with a new hybrid loss to provide a map, patch, and pixel‐level supervision. We also introduce a dataset called MFSD, with 11,601 images and 12,758 masked faces for masked face segmentation. Furthermore, we compare the performance of different cutting‐edge deep learning semantic segmentation models on the presented dataset. Experimental results on the MSFD dataset reveal that the suggested approach outperforms state‐of‐the‐art, algorithms with 97.623% accuracy, 93.814% IoU, and 96.817% F1‐score rate. Our dataset of masked faces with mask region labels and source code will be available online.

E2F-Net: Eyes-to-face inpainting via StyleGAN latent space

Face inpainting, the technique of restoring missing or damaged regions in facial images, is pivotal for applications like face recognition in occluded scenarios and image analysis with poor-quality captures. This process not only needs to produce realistic visuals but also preserve individual identity characteristics. The aim of this paper is to inpaint a face given periocular region (eyes-to-face) through a proposed new Generative Adversarial Network (GAN)-based model called Eyes-to-Face Network (E2F-Net). The proposed approach extracts identity and non-identity features from the periocular region using two dedicated encoders have been used. The extracted features are then mapped to the latent space of a pre-trained StyleGAN generator to benefit from its state-of-the-art performance and its rich, diverse and expressive latent space without any additional training. We further improve the StyleGAN's output to find the optimal code in the latent space using a new optimization for GAN inversion technique. Our E2F-Net requires a minimum training process reducing the computational complexity as a secondary benefit. Through extensive experiments, we show that our method successfully reconstructs the whole face with high quality, surpassing current techniques, despite significantly less training and supervision efforts. We have generated seven eyes-to-face datasets based on well-known public face datasets for training and verifying our proposed methods. The code and datasets are publicly available.11https://github.com/fatemejamalii/E2F-Net

Low-Cost Training of Image-to-Image Diffusion Models with Incremental Learning and Task/Domain Adaptation

Diffusion models specialized in image-to-image translation tasks, like inpainting and colorization, have outperformed the state of the art, yet their computational requirements are exceptionally demanding. This study analyzes different strategies to train image-to-image diffusion models in a low-resource setting. The studied strategies include incremental learning and task/domain transfer learning. First, a base model for human face inpainting is trained from scratch with an incremental learning strategy. The resulting model achieves an FID score almost equivalent to that of its batch learning equivalent while significantly reducing the training time. Second, the base model is fine-tuned to perform a different task, image colorization, and, in a different domain, landscape images. The resulting colorization models showcase exceptional performances with a minimal number of training epochs. We examine the impact of different configurations and provide insights into the ability of image-to-image diffusion models for transfer learning across tasks and domains.

Recovery-Based Occluded Face Recognition by Identity-Guided Inpainting.

Occlusion in facial photos poses a significant challenge for machine detection and recognition. Consequently, occluded face recognition for camera-captured images has emerged as a prominent and widely discussed topic in computer vision. The present standard face recognition methods have achieved remarkable performance in unoccluded face recognition but performed poorly when directly applied to occluded face datasets. The main reason lies in the absence of identity cues caused by occlusions. Therefore, a direct idea of recovering the occluded areas through an inpainting model has been proposed. However, existing inpainting models based on an encoder-decoder structure are limited in preserving inherent identity information. To solve the problem, we propose ID-Inpainter, an identity-guided face inpainting model, which preserves the identity information to the greatest extent through a more accurate identity sampling strategy and a GAN-like fusing network. We conduct recognition experiments on the occluded face photographs from the LFW, CFP-FP, and AgeDB-30 datasets, and the results indicate that our method achieves state-of-the-art performance in identity-preserving inpainting, and dramatically improves the accuracy of normal recognizers in occluded face recognition.

Face inpainting based on edge confrontation combined with hierarchical gated convolution

Face inpainting based on edge confrontation combined with hierarchical gated convolution

Enhancing Face Recognition with Detachable Self-Supervised Bypass Networks.

Attributed to the development of deep networks and abundant data, automatic face recognition (FR) has quickly reached human-level capacity in the past few years. However, the FR problem is not perfectly solved in case of large poses and uncontrolled occlusions. In this paper, we propose a novel bypass enhanced representation learning (BERL) method to improve face recognition under unconstrained scenarios. The proposed method integrates self-supervised learning and supervised learning together by attaching two auxiliary bypasses, a 3D reconstruction bypass and a blind inpainting bypass, to assist robust feature learning for face recognition. Among them, the 3D reconstruction bypass enforces the face recognition network to encode pose independent 3D facial information, which enhances the robustness to various poses. The blind inpainting bypass enforces the face recognition network to capture more facial context information for face inpainting, which enhances the robustness to occlusions. The whole framework is trained in end-to-end manner with two self-supervised tasks above and the classic supervised face identification task. During inference, the two auxiliary bypasses can be detached from the face recognition network, avoiding any additional computational overhead. Extensive experimental results on various face recognition benchmarks show that, without any cost of extra annotations and computations, our method outperforms state-of-the-art methods. Moreover, the learnt representations can also well generalize to other face-related downstream tasks such as the facial attribute recognition with limited labeled data.

Reference-Guided Large-Scale Face Inpainting With Identity and Texture Control

Face inpainting aims at plausibly predicting missing pixels of face images within a corrupted region. Most existing methods rely on generative models learning a face image distribution from a big dataset, which produces uncontrollable results, especially with large-scale missing regions. To introduce strong control for face inpainting, we propose a novel reference-guided face inpainting method that fills the large-scale missing region with identity and texture control guided by a reference face image. However, generating high-quality results under imposing two control signals is challenging. To tackle such difficulty, we propose a dual control one-stage framework that decouples the reference image into two levels for flexible control: High-level identity information and low-level texture information, where the identity information figures out the shape of the face and the texture information depicts the component-aware texture. To synthesize high-quality results, we design two novel modules referred to as Half-AdaIN and Component-Wise Style Injector (CWSI) to inject the two kinds of control information into the inpainting processing. Our method produces realistic results with identity and texture control faithful to reference images. To the best of our knowledge, it is the first work to concurrently apply identity and component-level controls in face inpainting to promise more precise and controllable results.

DEAR-GAN: Degradation-Aware Face Restoration With GAN Prior

With the development of generative adversarial networks (GANs), recent face restoration (FR) methods often utilize pre-trained GAN models ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e</i> ., StyleGAN2) as prior to generate rich details. However, these methods usually struggle to balance realness and fidelity when facing various degradation levels. In this paper, we propose a novel DEgradation-Aware Restoration network with GAN prior, dubbed DEAR-GAN, for FR tasks by explicitly learning the degradation representations (DR) to adapt to various degradation. Specifically, an unsupervised degradation representation learning (UDRL) strategy is first developed to extract DR of the input degraded images. Then, a degradation-aware feature interpolation (DAFI) module is proposed to dynamically fuse the two types of informative features ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e</i> ., features from degraded images and features from GAN prior network) with flexible adaption to various degradation based on DR. Extensive experiments show that our proposed DEAR-GAN outperforms the state-of-the-art methods for face restoration under multiple degradation and face super-resolution, and demonstrate the effectiveness of feature interpolation, which can be extended to face inpainting to achieve excellent results.

Pluralistic Face Inpainting With Transformation of Attribute Information

Most face-inpainting methods perform well in face repair. However, these methods can only complete a single face image per input. Although existing various image-inpainting methods can achieve pluralistic image inpainting, they typically produce faces with distorted structures or the same texture. To resolve these shortcomings and achieve high-quality diverse face inpainting, we propose PFTANet, a two-stage pluralistic face-inpainting network that transforms attribute information. In the first stage, the face-parsing network is fine-tuned to obtain semantic facial region information. In the second stage, a generator consisting of SNBlock, CF_ShiftBlocks, and CF_MergeBlock, which ensures that high-quality pluralistic face results are generated, is used. Specifically, CF_ShiftBlocks completes pluralistic face generation by transforming the attribute information from the conditional face extracted by the attribute extractor and ensuring the consistency of the attribute information between the conditional and generated faces. CF_MergeBlock ensures structural consistency between the masked and background regions of the generated face using facial region semantic information. A multi-patch discriminator is used to enhance facial detail generation. Experimental results for the CelebA and CelebA-HQ datasets indicated that PFTANet achieved pluralistic and visually realistic face inpainting.

FT-TDR: Frequency-Guided Transformer and Top-Down Refinement Network for Blind Face Inpainting

Blind face inpainting refers to the task of reconstructing visual contents without explicitly indicating the corrupted regions in a face image. Inherently, this task faces two challenges: (1) how to detect various mask patterns of different shapes and contents; (2) how to restore visually plausible and pleasing contents in the masked regions. In this paper, we propose a novel two-stage blind face inpainting method named Frequency-guided Transformer and Top-Down Refinement Network (FT-TDR) to tackle these challenges. Specifically, we first use a transformer-based network to detect the corrupted regions to be inpainted as masks by modeling the relation among different patches. For improved detection results, we also exploit the frequency modality as complementary information and capture the local contextual incoherence to enhance boundary consistency. Then a top-down refinement network is proposed to hierarchically restore features at different levels and generate contents that are semantically consistent with the unmasked face regions. Extensive experiments demonstrate that our method outperforms current state-of-the-art blind and non-blind face inpainting methods qualitatively and quantitatively.

Deep Face Video Inpainting via UV Mapping.

This paper addresses the problem of face video inpainting. Existing video inpainting methods target primarily at natural scenes with repetitive patterns. They do not make use of any prior knowledge of the face to help retrieve correspondences for the corrupted face. They therefore only achieve sub-optimal results, particularly for faces under large pose and expression variations where face components appear very differently across frames. In this paper, we propose a two-stage deep learning method for face video inpainting. We employ 3DMM as our 3D face prior to transform a face between the image space and the UV (texture) space. In Stage I, we perform face inpainting in the UV space. This helps to largely remove the influence of face poses and expressions and makes the learning task much easier with well aligned face features. We introduce a frame-wise attention module to fully exploit correspondences in neighboring frames to assist the inpainting task. In Stage II, we transform the inpainted face regions back to the image space and perform face video refinement that inpaints any background regions not covered in Stage I and also refines the inpainted face regions. Extensive experiments have been carried out which show our method can significantly outperform methods based merely on 2D information, especially for faces under large pose and expression variations. Project page: https://ywq.github.io/FVIP.

Generative face inpainting hashing for occluded face retrieval.

COVID-19 has resulted in a significant impact on individual lives, bringing a unique challenge for face retrieval under occlusion. In this paper, an occluded face retrieval method which consists of generator, discriminator, and deep hashing retrieval network is proposed for face retrieval in a large-scale face image dataset under variety of occlusion situations. In the proposed method, occluded face images are firstly reconstructed using a face inpainting model, in which the adversarial loss, reconstruction loss and hash bits loss are combined for training. With the trained model, hash codes of real face images and corresponding reconstructed face images are aimed to be as similar as possible. Then, a deep hashing retrieval network is used to generate compact similarity-preserving hashing codes using reconstructed face images for a better retrieval performance. Experimental results show that the proposed method can successfully generate the reconstructed face images under occlusion. Meanwhile, the proposed deep hashing retrieval network achieves better retrieval performance for occluded face retrieval than existing state-of-the-art deep hashing retrieval methods.

SwapInpaint: Identity-Specific Face Inpainting With Identity Swapping

As face editing scenarios have become popular, the face inpainting technique has become a hot topic. Although some existing methods can inpaint faces with preserved identity information, they fail to solve a more flexible inpainting problem that fills the “holes” with identity-specific content from other faces. In this work, we propose a disentangle and subject-agnostic framework that affects both full and partial-face inpainting with the guidance of a reference face image. The framework consists of an identity encoding module, a content inference module and a generative module. The identity encoding module extracts the identity embedding from the reference image, the content inference module learns to predict the content image, and the generative module integrates the content image and the reference identity embedding to generate the identity-specific inpainted result. To minimize the structure and style gap between the incomplete image and inpainted image, we use a double attribute loss to the generative module and a postprocess of blending operation to the swapped result. We compare our method with state-of-the-art works and demonstrate that our method achieves higher identity similarity and better structural correctness.