Facial identity recognition using StyleGAN3 inversion and improved tiny YOLOv7 model

Sad expressions during encoding enhance facial identity recognition in visual working memory in depression: Behavioural and electrophysiological evidence

This research sought to examine differences in the cross-modal facilitation effect of voice on facial identity recognition under mask occlusion for both oneself and others. Employing a facial recognition paradigm, we examined the influence of voice on facial identity recognition under static and dynamic mask occlusion through two eye-tracking experiments. The behavioral results from Experiments 1 and 2 indicate that mask occlusion interfered with recognition for both static and dynamic faces, with greater interference observed for others' faces than for self-faces. In addition, voice exerts cross-modal enhancement effects on faces, with greater enhancement observed for masked faces than for no mask. Furthermore, voice provides stronger enhancement for others' dynamic faces than for their self-dynamic faces. Eye-tracking data from both experiments revealed that the difference in dynamic facial recognition between self-faces and others' faces due to voice emerged in the early stages of dynamic facial recognition and persisted into later stages. However, regardless of whether they were in the early or late stages of static facial recognition, the facilitation effect of voice did not differ between themselves and others. This study revealed that the cross-modal facilitation of visual stimuli by voice is influenced by the self-advantage effect.

Face identity recognition in autism spectrum disorders: A review of behavioral studies

Impairment in emotion perception represents a fundamental feature of schizophrenia with important consequences in social functioning. A fundamental unresolved issue is the relationship between emotion perception and face perception. The aim of the present study was to examine whether facial identity recognition (Identity Discrimination) is a factor predicting facial emotion recognition in the context of the other factors, known as contributing to emotion perception, such as cognitive functions and symptoms. We enrolled 58 stable schizophrenic out-patients and 47 healthy subjects. Facial identity recognition and emotion perception were assessed with the Comprehensive Affect Testing System. Different multiple regression models with backward elimination were performed in order to discover the relation of each significant variable with emotion perception. In a regression including the six significant variables (age, positive symptomatology, Identity Discrimination, attentive functions, verbal memory-learning, executive functions) versus emotion processing, only attentive functions (standardised β = 0.264, p = 0.038) and Identity Discrimination (standardised β = 0.279, p = 0.029) reached a significant level. Two partial regressions were performed including five variables, one excluding attentive functions and the other excluding Identity Discrimination. When we excluded attentive functions, the only significant variable was Identity Discrimination (standardised β = 0.278, p = 0.032). When we excluded Identity Discrimination, both verbal memory-learning (standardised β = 0.261, p = 0.042) and executive functions (standardised β = 0.253, p = 0.048) were significant. Our results emphasised the role of face perception and attentional abilities on affect perception in schizophrenia. We additionally found a role of verbal memory-learning and executive functions on emotion perception. The relationship between those above-mentioned variables and emotion processing could have implications for cognitive rehabilitation.

BackgroundPeople with 22q11.2 deletion syndrome (22q11DS) have difficulty processing social information including facial identity and emotion processing. However, difficulties with visual and attentional processes may play a role in difficulties observed with these social cognitive skills.MethodsA cross-sectional study investigated visual perception and processing as well as facial processing abilities in a group of 49 children and adolescents with 22q11DS and 30 age and socio-economic status-matched healthy sibling controls using the Birmingham Object Recognition Battery and face processing sub-tests from the MRC face processing skills battery.ResultsThe 22q11DS group demonstrated poorer performance on all measures of visual perception and processing, with greatest impairment on perceptual processes relating to form perception as well as object recognition and memory. In addition, form perception was found to make a significant and unique contribution to higher order social-perceptual processing (face identity) in the 22q11DS group.ConclusionsThe findings indicate evidence for impaired visual perception and processing capabilities in 22q11DS. In turn, these were found to influence cognitive skills needed for social processes such as facial identity recognition in the children with 22q11DS.

The present study aims to explore the influence of facial emotional expressions on pre-scholars' identity recognition was analyzed using a two-alternative forced-choice matching task. A decrement was observed in children's performance with emotional faces compared with neutral faces, both when a happy emotional expression remained unchanged between the target face and the test faces and when the expression changed from happy to neutral or from neutral to happy between the target and the test faces (Experiment 1). Negative emotional expressions (i.e. fear and anger) also interfered with children's identity recognition (Experiment 2). Obtained evidence suggests that in preschool-age children, facial emotional expressions are processed in interaction with, rather than independently from, the encoding of facial identity information. The results are discussed in relationship with relevant research conducted with adults and children.

This paper introduces a new concept of LLE eigenface modelled by local linear embedding (LLE), and compares it with the traditional PCA eigenface from principle component analysis (PCA) on pose identity and face identity recognition through face classification. LLE eigenface is found outperforming PCA eigenface on the discrimination/recogntion of both face identity and pose identity. The superiority on face identity recognition is own to a more balanced energy distribution on LLE eigenfaces, while the superiority on pose identity recognition is due to the fact that LLE preserves a better local neighborhood of face images.KeywordsFace RecognitionFace ImageIndependent Component AnalysisLocally Linear EmbeddingFace IdentityThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

BackgroundThe integrity of face identity recognition (FIR) in prodromal Alzheimer’s disease (prAD) is inconsistent in the literature, and the cognitive processes underpinning a potential impairment remain elusive. The concepts of pattern separation (PS) and pattern completion (PC) offer an interesting insight in this respect, since efficient FIR implies that facial identities are stored in at least partly distinct representations (corresponding to a PS process) and to fill‐in the currently perceived facial pattern based on these mnemonic representations (corresponding to the PC process). The goal of our study was therefore to investigate FIR in prAD patients with a PC and PS paradigm. In addition, we aimed to verify that prAD patients were able to individuate faces in a task that does not involve any long‐term memory component.MethodThirty‐two healthy older participants and 17 prAD patients were familiarized with 40 facial identities, through a picture‐matching task (Fig. 1.A). Participants then performed a forced‐choice recognition task aiming at assessing and strengthening the memory encoding of these 40 faces (Fig. 1.B). They then underwent separate yes/no recognition tasks involving parametrically controlled blurred or morphed faces (Fig. 2), designed to solicit PC and PS processes respectively (Fig. 1.C‐D). Finally, in two separate discrimination tasks (Fig. 1.E‐F), participants were instructed to determine whether two simultaneously displayed faces were identical or not. These discrimination tasks included the blurred and morphed faces displayed in the PC and PS recognition tasks respectively.ResultsprAD patients showed significantly lower performance than controls in both the PC and PS recognition tasks (Fig. 3). However, this group difference was not significant when the recognition accuracy index for the non‐manipulated faces (not blurred nor morphed) displayed in these tasks was included as a covariate. In addition, group differences were unexpectedly found for the familiarization and discrimination tasks.ConclusionOur study failed to highlight specific PC or PS deficits in prAD patients. However, it showed that prAD patients performed poorly on tasks that required comparing and matching simultaneously presented facial patterns, in the absence of any long‐term memory component. Future research should investigate whether this impairment could represent an early cognitive marker of AD that could be easily assessed in clinical practice.

Introduction: Prosopagnosia is a striking neuropsychological condition characterized by severe deficits in facial identity recognition despite normal early visual processing and high-level cognition. Individuals experiencing prosopagnosia have great difficulty recognizing friends, family, and even their own face. Reports of prosopagnosia in the neurological literature date back to the mid-19th century (Wigan, 1844), but Hoff and Potzl (1937) and then Bodamer (in 1947; see Ellis & Florence, 1990) were the first to suggest that selective face recognition deficits indicate that face recognition involves different mechanisms from those used for other types of visual recognition. These early studies of prosopagnosia inspired further investigations of this issue using a wide range of methods (Gross, Rocha-Miranda, & Bender, 1972; Sergent, Ohta, & MacDonald, 1992; Yin, 1969), but despite hundreds of relevant studies, debate about the mechanisms of face recognition continues.

Integration of identity and emotion information in faces: fMRI evidence

<p>The eyes of other people subserve two core functions in human social cognition: gaze perception and face identity recognition. This thesis reports two psychophysical studies that examine the degree of functional specificity between eye gaze processing and face identity processing by testing if various aspects of gaze processing are intact in people with developmental prosopagnosia (DP) – the lifelong inability to recognise face identity. Study 1 investigates spatial integration in eye gaze perception using two tasks. DP and control participants completed one task that measured perception of gaze direction from the two eyes and another that measured the Wollaston illusion (whereby perceived eye gaze is pulled by head rotation; requiring the integration of eye and head direction). Study 2 investigates temporal integration in eye gaze perception using two tasks. The first task measured adaptation effects in eye gaze perception, which reflects sensitivity to gaze direction and its sensory representations. The second task measured serial dependence in gaze perception, which reflects temporal integration of gaze direction and its perceptual stability. Despite their deficits in recognising face identity, DP participants showed normal gaze processing across all studies. These results demonstrate the functional specificity of gaze processing and imply that gaze perception is carried out by dedicated mechanisms not used for processing identity. Our findings align with models of face processing that posit distinct pathways for gaze and identity analysis, and further clarify the selectivity of face processing dysfunctions in developmental prosopagnosia.</p>

Face recognition is the process of identifying and verifying a person from image or video. A significant amount of contributions have already been made by the researchers in this field of face identity techniques and recognition. In this paper, we further explored and investigated the evolution of face recognition methods from low-level features extracted from global features such as PCA, eigen face and SVM-based methods to high-level features extracted from deep learning models such as DeepFace and VGGFace. We also discussed the challenges such as illumination and pose variation and available standard data sets such as LFW and Yale data set in the field of face recognition.KeywordsFace recognitionGlobal methodsDeep learning methodsPCASVMDeepFaceVGGFace

We present the novel concept of Controllable Face Privacy. Existing methods that alter face images to conceal identity inadvertently also destroy other facial attributes such as gender, race or age. This all-or-nothing approach is too harsh. Instead, we propose a flexible method that can independently control the amount of identity alteration while keeping unchanged other facial attributes. To achieve this flexibility, we apply a subspace decomposition onto our face encoding scheme, effectively decoupling facial attributes such as gender, race, age, and identity into mutually orthogonal subspaces, which in turn enables independent control of these attributes. Our method is thus useful for nuanced face de-identification, in which only facial identity is altered, but others, such gender, race and age, are retained. These altered face images protect identity privacy, and yet allow other computer vision analyses, such as gender detection, to proceed unimpeded. Controllable Face Privacy is therefore useful for reaping the benefits of surveillance cameras while preventing privacy abuse. Our proposal also permits privacy to be applied not just to identity, but also to other facial attributes as well. Furthermore, privacy-protection mechanisms, such as k-anonymity, L-diversity, and t-closeness, may be readily incorporated into our method. Extensive experiments with a commercial facial analysis software show that our alteration method is indeed effective.

Face recognition systems which perform in unconstrained environment needs to cope with vary input resolutions. Low resolution input images are often quite problematic. This paper describes the improvement obtained on facial expression recognition and identity recognition when applying resolution upscaling methods based on trained filter techniques. In this paper, we explore the method proposed by Kondo et al. Kondo's methode has already been applied in display technology successfully, but hasn't been studied for recognition tasks before. In addition, we examine the influence on the recognition performance of the type of material exploited to train the upscaling filters, by selecting both generic videos and face images as samples. We then analyze how the size of the training set used in generating the upscaling filters influence the recognition rates. We show that adopting faces as training material produces upscaling filters capable of both higher recognition performance and much reduced requirements on the training size.

Facial attribute recognition is conventionally computed from a single image. In practice, each subject may have multiple face images. Taking the eye size as an example, it should not change, but it may have different estimation in multiple images, which would make a negative impact on face recognition. Thus, how to compute these attributes corresponding to each subject rather than each single image is a profound work. To address this question, we deploy deep training for facial attributes prediction, and we explore the inconsistency issue among the attributes computed from each single image. Then, we develop two approaches to address the inconsistency issue. Experimental results show that the proposed methods can handle facial attribute estimation on either multiple still images or video frames, and can correct the incorrectly annotated labels. The experiments are conducted on two large public databases with annotations of facial attributes.