Face Search Research Articles

Searching for Better Quality and A Shorter Queue

Customers looking for service providers often face search frictions and have to trade off quality and availability. To understand customers' search behavior when they are confronted with a large collection of vertically differentiated, congested service providers, we build a model in which arriving customers conduct costly sequential search to resolve uncertainty about service providers' quality and queue length and select one according to an optimal stopping rule. Customers search due, in part, to variations in waiting time across service providers, which, in turn, are determined by the search behavior of customers. Thus, an equilibrium emerges. We characterize customers' equilibrium search/join behavior in a mean field model as the number of service providers grows large. We find that reducing either the search cost or customer arrival rate may not improve customer welfare and may strictly increase the average waiting time in the system as customers substitute toward high quality service providers. Moreover, with lower search costs, the improved quality obtained by customers may not make up for the prolonged wait, therefore degrading the average search reward. We discuss policy implications of our results in the context of public surgical waits.

An Iterative Approach for EEG-Based Rapid Face Search: A Refined Retrieval by Brain Computer Interfaces

Recent face recognition techniques have achieved remarkable successes in fast face retrieval on huge image datasets. But the performance is still limited when large illumination, pose, and facial expression variations are presented. In contrast, the human brain has powerful cognitive capability to recognize faces and demonstrates robustness across viewpoints, lighting conditions, even in the presence of partial occlusion. This paper proposes a closed-loop face retrieval system that combines the state-of-the-art face recognition method with the powerful cognitive function of the human brain illustrated in electroencephalography signals. The system starts with a random face image and outputs the ranking of all of the images in the database according to their similarity to the target individual. At each iteration, the single trial event related potentials (ERP) detector scores the user’s interest in rapid serial visual presentation paradigm, where the presented images are selected from the computer face recognition module. When the system converges, the ERP detector further refines the lower ranking to achieve better performance. In total, 10 subjects participated in the experiment, exploring a database containing 1,854 images of 46 celebrities. Our approach outperforms existing methods with better average precision, indicating human cognitive ability complements computer face recognition and contributes to better face retrieval.

Selection of Viola–Jones algorithm parameters for specific conditions

In this paper, we consider the face search in images by the Viola-Jones method. The method operation is controlled by a number of parameters affecting the balance between the numbers of false operations and missed faces, as well as the method speed. Selection of parameters for practical problems is rather laborious, since an optimal choice requires many experiments for a large data set. The objective of this study is to obtain a priori information on the effect of parameters on the method operation, the use of which would make it possible to efficiently select parameters without the need of preliminary collection of a test data set.

Face Classification and Recognition Based on Contour and Local Texture Feature

In order to achieve fast and accurate face recognition, firstly, this paper extracts face contour lines based on face contour feature to achieve treat a human face shape classification, thereby reducing the measured face search range. Secondly, using the method of Gabor wavelet to extract local texture on face when range was approximately determined, then matching with the face library whose range has been narrowed. Experiments show that this parallel approach not only achieved rapid recognition, but also achieve a high recognition rate due to a combination of the two methods.

In search for the many faces of community resilience among LGBT individuals.

Many countries (e.g., Egypt, Russia, and Uganda) have very draconian laws against LGBT individuals. Despite facing such adversity or hostility many LGBT individuals do strive and maintain their sense of self-worth. How do they do it? The collection of papers in this special issue attempts to provide some answers to this question. I will attempt to analyze two overlapping issues in these papers, and how they might inform LGBT research using a resilience lens.

Searching for faces of different ages: Evidence for an experienced-based own-age detection advantage in adults.

Previous studies have shown that attention deployment in visual search tasks is modulated by face race and emotional expression, with a search asymmetry in favor of those faces that are less efficiently discriminated and recognized at the individual level (i.e., other-race faces and angry faces). Face age is another dimension affecting how faces are remembered, as it has been widely reported that young adults show significant deficits in recognizing other-age faces. By comparing adults' search efficiency for own- and other-age faces in a visual search task in which face age was the target feature we explored whether the mirror pattern of detection and recognition effects found for race biases generalizes to age biases, and whether search efficiency for adult and nonadult faces is modulated by experience accumulated with nonadult faces. Search efficiency was greater for adult faces than for infant (Experiment 1) or child faces (Experiment 2) in adults with limited experience with infants or children, whereas there was no sign of search asymmetry in preschool teachers who have had extensive recent experience with children (Experiment 2). Results indicate that the influence of age on attention deployment parallels the effects that this face attribute has on face recognition, and that both effects are experience-based.

Investigating attention in complex visual search

How we attend to and search for objects in the real world is influenced by a host of low-level and higher-level factors whose interactions are poorly understood. The vast majority of studies approach this issue by experimentally controlling one or two factors in isolation, often under conditions with limited ecological validity. We present a comprehensive regression framework, together with a matlab-implemented toolbox, which allows concurrent factors influencing saccade targeting to be more clearly distinguished. Based on the idea of gaze selection as a point process, the framework allows each putative factor to be modeled as a covariate in a generalized linear model, and its significance to be evaluated with model-based hypothesis testing. We apply this framework to visual search for faces as an example and demonstrate its power in detecting effects of eccentricity, inversion, task congruency, emotional expression, and serial fixation order on the targeting of gaze. Among other things, we find evidence for multiple goal-related and goal-independent processes that operate with distinct visuotopy and time course.

Searching for a talking face: the effect of degrading the auditory signal.

Previous research (e.g., McGurk & MacDonald, 1976) suggests that faces and voices are bound automatically, but recent evidence suggests that attention is involved in a task of searching for a talking face (Alsius & Soto-Faraco, 2011). We hypothesized that the processing demands of the stimuli may affect the amount of attentional resources required, and investigated what effect degrading the auditory stimulus had on the time taken to locate a talking face. Twenty participants were presented with between 2 and 4 faces articulating different sentences, and had to decide which of these faces matched the sentence that they heard. The results showed that in the least demanding auditory condition (clear speech in quiet), search times did not significantly increase when the number of faces increased. However, when speech was presented in background noise or was processed to simulate the information provided by a cochlear implant, search times increased as the number of faces increased. Thus, it seems that the amount of attentional resources required vary according to the processing demands of the auditory stimuli, and when processing load is increased then faces need to be individually attended to in order to complete the task. Based on these results we would expect cochlear-implant users to find the task of locating a talking face more attentionally demanding than normal hearing listeners.

Eye movements during visual search for emotional faces in individuals with chronic headache.

Attentional biases for pain-related information have been frequently reported in individuals with chronic pain. Recording of participants' eye movements provides a continuous measure of attention, although to date this methodology has received little use in research exploring attentional biases in chronic pain. The aim of the current investigation was to explore the specificity of attentional orienting bias using a novel visual search task while recording participant eye movement behaviours. This also allowed for the investigation of whether attentional biases for pain-related information exist in the presence of multiple stimuli competing for attention. Twenty-three participants with chronic headache and 24 pain-free, healthy control participants were engaged in a visual search task where pain, angry, happy and neutral faces were used as both target and distractor stimuli. While completing this task, participants' eye movements were recorded. Supporting the adopted hypothesis, participants with chronic headache, relative to healthy controls, demonstrated a significantly higher proportion of initial fixations to target pain expressions when the pain expressions were presented in displays containing neutral-distractor faces. No significant differences were found between groups in the time taken to fixate target pain expressions (localization time). Individuals with chronic headache show facilitated initial orienting towards pain expressions specifically when used as targets in a visual search task. This study adds to a growing body of research supporting the presence of pain-related attentional biases in chronic pain as assessed via different experimental paradigms, and shows biases to exist when multiple stimuli competing for attention are presented simultaneously.

Finding faces among faces: human faces are located more quickly and accurately than other primate and mammal faces

We tested the specificity of human face search efficiency by examining whether there is a broad window of detection for various face-like stimuli-human and animal faces-or whether own-species faces receive greater attentional allocation. We assessed the strength of the own-species face detection bias by testing whether human faces are located more efficiently than other animal faces, when presented among various other species' faces, in heterogeneous 16-, 36-, and 64-item arrays. Across all array sizes, we found that, controlling for distractor type, human faces were located faster and more accurately than primate and mammal faces, and that, controlling for target type, searches were faster when distractors were human faces compared to animal faces, revealing more efficient processing of human faces regardless of their role as targets or distractors (Experiment 1). Critically, these effects remained when searches were for specific species' faces (human, chimpanzee, otter), ruling out a category-level explanation (Experiment 2). Together, these results suggest that human faces may be processed more efficiently than animal faces, both when task-relevant (targets) and task-irrelevant (distractors), even in direct competition with other faces. These results suggest that there is not a broad window of detection for all face-like patterns but that human adults process own-species' faces more efficiently than other species' faces. Such own-species search efficiencies may arise through experience with own-species faces throughout development or may be privileged early in development, due to the evolutionary importance of conspecifics' faces.

Managing Congestion in Decentralized Matching Markets

Participants in matching markets face search and screening costs which prevent the market from clearing efficiently. In many settings, the rise of online matching platforms has dramatically reduced the cost of finding and contacting potential partners. While one might expect both sides of the market to benefit from reduced search costs, this is far from guaranteed. In particular, this change may force participants to screen more potential partners before finding one who is willing to accept their offer. Thus, for a fixed screening cost, reductions in search and application costs may actually decrease aggregate welfare.We illustrate this fact using a model in which agents apply and screen asynchronously. In our model, it is possible that an agent on one side of the market (an 'employer') identifies a suitable match on the other side (an 'applicant'), only to find that this applicant has already matched. We find that equilibrium is generically inefficient for both employers and applicants. Most notably, when application costs are sufficiently small, uncertainty about applicant availability may drive equilibrium employer welfare to zero.We consider a simple intervention available to the platform: limiting the visibility of applicants. We find that this intervention can significantly improve the welfare of agents on both sides of the market; applicants pay lower application costs, while employers are less likely to find that the applicants they screen have already matched. Somewhat counterintuitively, the benefits of showing fewer applicants to each employer are greatest in markets in which there is a shortage of applicants.

A collaborative approach for face verification and attributes refinement

Face verification which aims to determine whether two given faces refer to the same person, and human attribute learning with the goal of extracting predefined describable attributes from face images, are two fundamental issues in a variety of applications (e.g., face tagging, attribute based face search). While advances in computer vision domain have resulted in a series of techniques for each of the two tasks, such techniques are usually prone to errors due to the large variation of faces in pose, expression, illumination, occlusion, etc. Different from most prior related works which focus on the two tasks separately, in this paper, we explore their relationships and propose a collaborative approach allowing them to interact with each other to iteratively reduce errors and uncertainties. The interaction is embodies in two processes, one is that the results of face verification can be leveraged to refine attribute values utilizing the random-walk model, and the other is that the attribute values can also be employed to improve the face verification performance through R-LDA model. The two interactive processes will continue to iteratively improve the performance of the two tasks, until the relative stable results are achieved. Experimental results on the real-world photo collections demonstrate the effectiveness of the proposed approach.

Scanpath Generated by Cue-Driven Activation and Spatial Strategy: A Comparative Study

A comparative study of a cued face search task is presented in this paper. Human participants and a computer model carried out a task in which they were required to locate a color-cued target face. Human-generated eye fixations and scanpaths were compared with those generated by the computational model. Throughout the comparison, we considered the similarities and dissimilarities between the two systems’ performances. Their results show that the eye fixations in a valid cue search are highly correlated with the computer-generated fixation points in a valid cue search but not to those in random and invalid cue searches. Moreover, the comparison between human- and computer-generated scanpaths showed that the scanpath that links the fixation points is not randomly generated. Our results imply that eye movement is accomplished not only by cue-driven activation, but also by a spatial strategy.

Familiar and Unfamiliar Face Recognition in a Crowd

The purpose of the present study was to investigate whether the visual search process for familiar faces differs from that for unfamiliar faces. We used a single-target visual search task and recorded eye movements of participants during the task. We employed three different types of face familiarity: personally familiar faces (friends and teachers), famous faces, and unfamiliar faces. Participants had to search through arrays of faces for a target face. In each trial, a target face and three distractor faces were horizontally aligned and presented at the same time. In the personally familiar or famous condition, the target was a personally familiar face or a famous face, respectively, and the distractor faces were unfamiliar. In the unfamiliar condition, the target was an unfamiliar face and the distractor faces were personally familiar. The reaction times to identify the target demonstrated that the visual search for unfamiliar faces is slower than that for personally familiar faces and famous faces, but there was no significant difference between reaction times to recognize the personally familiar and the famous face targets. Additionally, the eye movement results in the unfamiliar face condition showed that an exhaustive search of the entire array occurred more frequently than a self-terminating search, but this was not true for both the personally familiar faces and famous faces conditions. These results suggest that the visual search process for familiar faces (personally familiar and famous) in a crowd differs from that for unfamiliar faces.

视觉工作记忆内容引导情绪面孔的视觉搜索

本文通过采用记忆任务和视觉搜索任务相结合的方式，以简笔画情绪面孔为材料，探讨了视觉工作记忆内容对情绪面孔视觉搜索的影响。结果发现：不同工作记忆内容条件下对情绪面孔搜索的反应时存在显著差异，对两种情绪面孔搜索的反应时在无效信息条件下显著慢于有效信息条件下。结果表明，工作记忆内容能够引导情绪面孔的视觉搜索,实验结果拓展了偏向竞争理论模型的适用范围。 With the combined memory task and visual search paradigm, this study aimed to discuss the influence of contents of visual working memory to visual search of emotional faces. The results showed that the effect of the contents of visual working memory was significant. And the RTs were significantly longer in the invalid condition than in the valid condition for not only happy face but also angry face. The results suggested that the information which was maintained in visual working memory could affect the visual search efficiency of emotional faces. The findings broaden biased competition model.

Visual search efficiency is greater for human faces compared to animal faces.

The Animate Monitoring Hypothesis proposes that humans and animals were the most important categories of visual stimuli for ancestral humans to monitor, as they presented important challenges and opportunities for survival and reproduction; however, it remains unknown whether animal faces are located as efficiently as human faces. We tested this hypothesis by examining whether human, primate, and mammal faces elicit similar searches, or whether human faces are privileged. In the first three experiments, participants located a target (human, primate, or mammal face) among distractors (non-face objects). We found fixations on human faces were faster and more accurate than fixations on primate faces, even when controlling for search category specificity. A final experiment revealed that, even when task-irrelevant, human faces slowed searches for non-faces, suggesting some bottom-up processing may be responsible for the human face search efficiency advantage.

Association and dissociation between detection and discrimination of objects of expertise: Evidence from visual search

Expertise in face recognition is characterized by high proficiency in distinguishing between individual faces. However, faces also enjoy an advantage at the early stage of basic-level detection, as demonstrated by efficient visual search for faces among nonface objects. In the present study, we asked (1) whether the face advantage in detection is a unique signature of face expertise, or whether it generalizes to other objects of expertise, and (2) whether expertise in face detection is intrinsically linked to expertise in face individuation. We compared how groups with varying degrees of object and face expertise (typical adults, developmental prosopagnosics [DP], and car experts) search for objects within and outside their domains of expertise (faces, cars, airplanes, and butterflies) among a variable set of object distractors. Across all three groups, search efficiency (indexed by reaction time slopes) was higher for faces and airplanes than for cars and butterflies. Notably, the search slope for car targets was considerably shallower in the car experts than in nonexperts. Although the mean face slope was slightly steeper among the DPs than in the other two groups, most of the DPs' search slopes were well within the normative range. This pattern of results suggests that expertise in object detection is indeed associated with expertise at the subordinate level, that it is not specific to faces, and that the two types of expertise are distinct facilities. We discuss the potential role of experience in bridging between low-level discriminative features and high-level naturalistic categories.

The Brain Potential Features of Human Face Recognition during Working Memory

As one of the most important skills in human's daily life, face recognition can help us to promote social interactions and adapt to the environment. Previous ERP studies showed that face recognition could induce N170(a negative potential peaking at about 170ms) over lateral occipito-temporal electrodes, some researchers argued that N170 might reflect the early structural encoding process. It is widely recognized that N170 possesses right hemisphere advantage, which suggests that the right hemisphere of brain is mainly responsible for structural encoding process. Searching for a specific face(the target) in the crowd and distinguishing it from other's faces(distractors) are related to working memory which refers to the limited resource system for temporary storage and information processing. Since it is an important function of working memory to intentionally hold an item in mind for current use, the relevant research on working memory is more than 30years. But the brain potential features of human face recognition during working memory as well as the brain responses to matching faces affected by prior learning remain unclear. Therefore, this article employs event-related potentials(ERPs) technology to explore the essence of face recognition N170 effect and the brain potential features of face recognition during working memory. The influence of prior learning on tracking and discrimination processes is also examined, which depends on the working memory status of a face.The delayed match-to-sample task paradigm was used in this study with pictures of faces as stimuli. 16college students(eight male and eight female, Mean age = 20.31 years) participated in the experiment and performed sample target match tasks during working memory, tracking targets(familiar and new targets) and excluding the distractors(familiar and new distractors).The ERP results revealed that N170 was induced over lateral occipito-temporal electrodes when participants recognized the targets and distractors, and N170 amplitudes of targets and distractors at the same electrode had no significant differences. Whereas, there was a significant difference between P7 and P8 when different electrodes were compared with; Both the targets and distractors evoked a late positive ERP component peaking around 450ms(P300) during the working memory process. The ERPs of targets and distractors showed dissociation as early as 250ms when tracking new or studied targets respectively. The targets evoked larger and more positive ERPs than the distractors, and new targets evoked a more positive ERP waveform than the studied ones. The typical old/new effect was observed between ERPs of studied and new distractors about 250~650 ms in prefrontal area.The present study results suggest that face recognition N170 effect reflects the whole processing of perception, and there is a right hemisphere superiority of N170 — especially a right temporal region superiority; Moreover, our current findings demonstrated that prior learning affects brain responses to matching faces during a working memory task.

Perceptual load effects on processing distractor faces indicate face-specific capacity limits

The claim that face perception is mediated by a specialized “face module” that proceeds automatically, independently of attention (e.g., Kanwisher, 2000) can be reconciled with load theory claims that visual perception has limited capacity (e.g., Lavie, 1995) by hypothesizing that face perception has face-specific capacity limits. We tested this hypothesis by comparing the effects of face and nonface perceptual load on distractor face processing. Participants searched a central array of either faces or letter strings for a pop star versus politician's face or name and made speeded classification responses. Perceptual load was varied through the relevant search set size. Response competition effects from a category-congruent or -incongruent peripheral distractor face were eliminated with more than two faces in the face search task, but were unaffected by perceptual load in the name search task. These results support the hypothesis that face perception has face-specific capacity limits and resolve apparent discrepancies in previous research.

Hybrid Algorithm for Face Recognition Using an Evolutionary Software Engineering

The principal aim was the construction of a face recognition system in order to be implemented in the service robot Donaxi, delimited by the Who is who test which is part of the RoboCups tests set, using an evolutionary development strategy of triple iterations. A two phase hybrid algorithm was developed, the first phase aim was the face detection using the Haar classifiers for face search in an image and the second phase is based on a decision tree whereby the faces characteristics were evaluated by the comparison techniques of phase correlation and histogram comparison. The needed characteristics were identified in order to develop this work as a software engineering project which allowed the algorithm construction and implementation through an evolutionary approach and a personal development process. The evolutionary strategy allowed the prototyping development with functionality and the tracking of the final system construction. A three iterations total was realized during which the needed metrics were registered (time, defects and sizes). The final analysis of results (algorithm and methods) allowed concluding and visualizing the employment advantages of a software engineering formal technique for research and robotics projects realization when improving estimations and software production quality.