1,720,961 research outputs found
Image Similarity between Masked and Unmasked Face for Consumer Electronics Applications
No full textFace recognition has become essential as a convenient biometric-based solution for a plethora of different consumer electronics applications, including access control systems, intelligent environments, smartphone authentication systems and so on. Early in 2020, the COVID-19 pandemic caused the widespread use of face masks, which become essential for containing the outbreak. The masks cause a visible alteration in facial appearance, covering almost the 50% of the human face. In this work, an image similarity technique is applied to assess the difference between two images of the same face wearing or not wearing a face mask. Cosine Similarity measure-based Algorithm (CSA) was used to objectively infer the difficulties that modern facial recognition algorithms, based on deep learning techniques, encounter when dealing with a masked face
Gender classification on 2D human skeleton
No full textSoft bimetrics has become a trending research topic over the past decade. In last years the increase of new technologies such as the wearable camera devices has introduced a new challenge into the gender classification problem. In this sense the ability to classify the gender not by an image but by the 2D estimated skeleton points is considered in this paper. Our experiments show that the human gender can be classified just considering the pose information provided by the body pose information. The proposed method have shown a remarkable performance on a dataset where subjects and camera are in movement
Inflated 3D ConvNet context analysis for violence detection
No full textAccording to the Wall Street Journal, one billion surveillance cameras will be deployed around the world by 2021. This amount of information can be hardly managed by humans. Using a Inflated 3D ConvNet as backbone, this paper introduces a novel automatic violence detection approach that outperforms state-of-the-art existing proposals. Most of those proposals consider a pre-processing step to only focus on some regions of interest in the scene, i.e., those actually containing a human subject. In this regard, this paper also reports the results of an extensive analysis on whether and how the context can affect or not the adopted classifier performance. The experiments show that context-free footage yields substantial deterioration of the classifier performance (2% to 5%) on publicly available datasets. However, they also demonstrate that performance stabilizes in context-free settings, no matter the level of context restriction applied. Finally, a cross-dataset experiment investigates the generalizability of results obtained in a single-collection experiment (same dataset used for training and testing) to cross-collection settings (different datasets used for training and testing)
Exploring biometric domain adaptation in human action recognition models for unconstrained environments
No full textIn conventional machine learning (ML), a fundamental assumption is that the training and test sets share identical feature distributions, a reasonable premise drawn from the same dataset. However, real-world scenarios often defy this assumption, as data may originate from diverse sources, causing disparities between training and test data distributions. This leads to a domain shift, where variations emerge between the source and target domains. This study delves into human action recognition (HAR) models within an unconstrained, real-world setting, scrutinizing the impact of input data variations related to contextual information and video encoding. The objective is to highlight the intricacies of model performance and interpretability in this context. Additionally, the study explores the domain adaptability of HAR models, specifically focusing on their potential for re-identifying individuals within uncontrolled environments. The experiments involve seven pre-trained backbone models and introduce a novel analytical approach by linking domain-related (HAR) and domain-unrelated (re-identification (re-ID)) tasks. Two key analyses addressing contextual information and encoding strategies reveal that maintaining the same encoding approach during training results in high task correlation while incorporating richer contextual information enhances performance. A notable outcome of this study is the comprehensive evaluation of a novel transformer-based architecture driven by a HAR backbone, which achieves a robust re-ID performance superior to state-of-the-art (SOTA). However, it faces challenges when other encoding schemes are applied, highlighting the role of the HAR classifier in performance variations
Inflated 3D ConvNet context analysis for violence detection
Full text linkAccording to the Wall Street Journal, one billion surveillance cameras will be deployed around the world by 2021. This amount of information can be hardly managed by humans. Using a Inflated 3D ConvNet as backbone, this paper introduces a novel automatic violence detection approach that outperforms state-of-the-art existing proposals. Most of those proposals consider a pre-processing step to only focus on some regions of interest in the scene, i.e., those actually containing a human subject. In this regard, this paper also reports the results of an extensive analysis on whether and how the context can affect or not the adopted classifier performance. The experiments show that context-free footage yields substantial deterioration of the classifier performance (2% to 5%) on publicly available datasets. However, they also demonstrate that performance stabilizes in context-free settings, no matter the level of context restriction applied. Finally, a cross-dataset experiment investigates the generalizability of results obtained in a single-collection experiment (same dataset used for training and testing) to cross-collection settings (different datasets used for training and testing)
Gotcha-I: A Multiview Human Videos Dataset
Full text linkThe growing need of security in large open spaces led to the need to use video capture of people in different context and illumination and with multiple biometric traits as head pose, body gait, eyes, nose, mouth, and further more. All these traits are useful for a multibiometric identification or a person re-identification in a video surveillance context. Body Worn Cameras (BWCs) are used by the police of different countries all around the word and their use is growing significantly. This raises the need to develop new recognition methods that consider multibiometric traits on person re-identification. The purpose of this work is to present a new video dataset called Gotcha-I. This dataset has been obtained using more mobile cameras to adhere to the data of BWCs. The dataset includes videos from 62 subjects in indoor and outdoor environments to address both security and surveillance problem. During these videos, subjects may have a different behavior in videos such as freely, path, upstairs, avoid the camera. The dataset is composed by 493 videos including a set of 180° videos for each face of the subjects in the dataset. Furthermore, there are already processed data, such as: the 3D model of the face of each subject with all the poses of the head in pitch, yaw and roll; and the body keypoint coordinates of the gait for each video frame. It’s also shown an application of gender recognition performed on Gotcha-I, confirming the usefulness and innovativeness of the proposed dataset
Gait Analysis for Gender Classification in Forensics
Full text linkGender Classification (GC) is a natural ability that belongs to the human beings. Recent improvements in computer vision provide the possibility to extract information for different classification/recognition purposes. Gender is a soft biometrics useful in video surveillance, especially in uncontrolled contexts such as low-light environments, with arbitrary poses, facial expressions, occlusions and motion blur. In this work we present a methodology for the construction of a gait analyzer. The methodology is divided into three major steps: (1) data extraction, where body keypoints are extracted from video sequences; (2) feature creation, where body features are constructed using body keypoints; and (3) classifier selection when such data are used to train four different classifiers in order to determine the one that best performs. The results are analyzed on the dataset Gotcha, characterized by user and camera either in motion
Going Beyond Counting First Authors in Author Co-citation Analysis
Full text linkThe present study examines one of the fundamental aspects of author co-citation analysis (ACA) - the way co-citation
counts are defined. Co-citation counting provides the data on which all subsequent statistical analyses and mappings
are based, and we compare ACA results based on two different types of co-citation counting - the traditional type that
only counts the first one among a cited work's authors on the one hand and a non-traditional type that takes into
account the first 5 authors of a cited work on the other hand. Results indicate that the picture produced through this non-traditional author co-citation counting contains more coherent author groups and is therefore considerably clearer. However, this picture represents fewer specialties in the research field being studied than that produced through the traditional first-author co-citation counting when the same number of top-ranked authors is selected and analyzed. Reasons for these effects are discussed
Variations on the Author
Full text link“Variations on the Author” discusses two of Eduardo Coutinho’s recent films (Um Dia na Vida, from 2010, and Últimas Conversas, posthumously released in 2015) and their contribution to the general question of documentary authorship. The director’s filmography is characterized by a consistent yet self-effacing form of authorial self-inscription: Coutinho often features as an interviewer that rather than express opinions propels discourses; an interviewer that is good at listening. This mode of self-inscription characterizes him as an author who is not expressive but who is nonetheless markedly present on the screen. In Um Dia na Vida, however, Coutinho is completely absent form the image, while Últimas Conversas, on the contrary, includes a confessional prologue that moves the director from the margins to the center of his films. This article examines the ways in which these works stand out in the filmography of a director who offers new insights into the notion of cinematic authorship
- …
