Showing papers by "Terrance E. Boult published in 2013"

Toward Open Set Recognition

Abstract: To date, almost all experimental evaluations of machine learning-based recognition algorithms in computer vision have taken the form of “closed set” recognition, whereby all testing classes are known at training time. A more realistic scenario for vision applications is “open set” recognition, where incomplete knowledge of the world is present at training time, and unknown classes can be submitted to an algorithm during testing. This paper explores the nature of open set recognition and formalizes its definition as a constrained minimization problem. The open set recognition problem is not well addressed by existing algorithms because it requires strong generalization. As a step toward a solution, we introduce a novel “1-vs-set machine,” which sculpts a decision space from the marginal distances of a 1-class or binary SVM with a linear kernel. This methodology applies to several different applications in computer vision where open set recognition is a challenging problem, including object recognition and face verification. We consider both in this work, with large scale cross-dataset experiments performed over the Caltech 256 and ImageNet sets, as well as face matching experiments performed over the Labeled Faces in the Wild set. The experiments highlight the effectiveness of machines adapted for open set evaluation compared to existing 1-class and binary SVMs for the same tasks.

Method and system for authenticating remote users

Abstract: A user of a mobile device can be authenticated based on multiple factors including biometric data of the user. During an enrollment process of the user, an encryption key is sent to the mobile device via a message. The encryption key is recovered from the message and used to encrypt communications between the mobile device and a server. Biometric data is collected from the user and sent to the server for computing a biometric model (e.g., a voice model, etc.) of the user for later use in authentication. An encrypted biometric model is stored only in the mobile device and the encrypted biometric model is sent to the server for authentication of the user. For authentication, various information including an identification of the mobile device, responses to challenge questions, biometric data including the biometric model, etc. are used at the server.

Animal recognition in the Mojave Desert: Vision tools for field biologists

Abstract: The outreach of computer vision to non-traditional areas has enormous potential to enable new ways of solving real world problems. One such problem is how to incorporate technology in the effort to protect endangered and threatened species in the wild. This paper presents a snapshot of our interdisciplinary team's ongoing work in the Mojave Desert to build vision tools for field biologists to study the currently threatened Desert Tortoise and Mohave Ground Squirrel. Animal population studies in natural habitats present new recognition challenges for computer vision, where open set testing and access to just limited computing resources lead us to algorithms that diverge from common practices. We introduce a novel algorithm for animal classification that addresses the open set nature of this problem and is suitable for implementation on a smartphone. Further, we look at a simple model for object recognition applied to the problem of individual species identification. A thorough experimental analysis is provided for real field data collected in the Mojave desert.

Secure voice-based authentication for mobile devices: vaulted voice verification

Abstract: As the use of biometrics becomes more wide-spread, the privacy concerns that stem from the use of biometrics are becoming more apparent. As the usage of mobile devices grows, so does the desire to implement biometric identification into such devices. A large majority of mobile devices being used are mobile phones. While work is being done to implement different types of biometrics into mobile phones, such as photo based biometrics, voice is a more natural choice. The idea of voice as a biometric identifier has been around a long time. One of the major concerns with using voice as an identifier is the instability of voice. We have developed a protocol that addresses those instabilities and preserves privacy. This paper describes a novel protocol that allows a user to authenticate using voice on a mobile/remote device without compromising their privacy. We first discuss the Vaulted Verification protocol, which has recently been introduced in research literature, and then describe its limitations. We then introduce a novel adaptation and extension of the Vaulted Verification protocol to voice, dubbed Vaulted Voice Verification ( V 3 ). Following that we show a performance evaluation and then conclude with a discussion of security and future work.

Large scale unconstrained open set face database

Abstract: This paper addresses large scale, unconstrained, open set face recognition, which exhibits the properties of operational face recognition scenarios. Most of the existing face recognition databases have been designed under controlled conditions or have been constructed from the images collected from the web. Face images collected from the web are less constrained than a mug-shot like collection. However, they lack information about the imaging conditions and have no operational paradigm. In either case, most of the databases and evaluation algorithms have taken the form of “closed set” recognition, in which all testing classes are assumed to be known at training time. A more realistic scenario in face recognition is an “open set,” where limited knowledge is available at training time and unknown classes can be present at test time. The database we provide supports the open set paradigm, which more closely mimics actual usage than classic closed set testing. The database also exhibits the natural variability among the face images such as pose, illumination, scale, expressions, occlusion, etc. Our goal is to provide around 100,000 images of more than 1,000 people. Also, with this paper, we release part 1 of the database, which consists of 6,337 images from 308 subjects. The paper discusses the details of the database followed by the challenges and results of baseline algorithms.

Good Recognition is Non-Metric

Abstract: Recognition is the fundamental task of visual cognition, yet how to formalize the general recognition problem for computer vision remains an open issue. The problem is sometimes reduced to the simplest case of recognizing matching pairs, often structured to allow for metric constraints. However, visual recognition is broader than just pair matching -- especially when we consider multi-class training data and large sets of features in a learning context. What we learn and how we learn it has important implications for effective algorithms. In this paper, we reconsider the assumption of recognition as a pair matching test, and introduce a new formal definition that captures the broader context of the problem. Through a meta-analysis and an experimental assessment of the top algorithms on popular data sets, we gain a sense of how often metric properties are violated by good recognition algorithms. By studying these violations, useful insights come to light: we make the case that locally metric algorithms should leverage outside information to solve the general recognition problem.

Voice authentication using short phrases: Examining accuracy, security and privacy issues

Abstract: This paper examines a novel security model for voice biometrics that decomposes the overall problem into bits of “biometric identity security,” bits of “knowledge security,” and bits of “traditional encryption security.” This is the first paper to examine balancing security gained from text-dependent and text-independent voice biometrics under this model. Our formulation allows for text-dependent voice biometrics to address both what you know and who you are. A text-independent component is added to defeat replay attacks. Further, we experimentally examine an extension of the recently introduced Vaulted Voice Verification protocol and the security tradeoffs of adding these elements. We show that by mixing text-dependent with text-independent voice verification and by expanding the challenge-response protocol, Vaulted Voice Verification can preserve privacy while addressing the problematic issues of voice as a remote/mobile biometric identifier. The resulting model supports both authentication and key release with the matching taking place client side, where a mobile device may be used. This novel security model addresses a real and crucial problem: that of security on a mobile device.

With vaulted voice verification my voice is my key

Abstract: Those who handle sensitive information from time to time need a device that can communicate securely. They also need the ability to verify the recipient of the information. For such secure communication to take place, they must securely exchange a key, often with someone they do not already know. Biometrics have been gaining widespread adoption in an effort to verify the end users identity. We extend this to key exchange. Vaulted Voice Verification, a recently introduced voice-based biometric protocol, has been shown to securely and remotely verify a user while also maintaining the privacy of the user. However Vaulted Voice Verification as originally proposed was not well suited for the exchange of larger keys. We present an index-based Vaulted Voice Verification which significantly reduces communication overhead and allows the transmission of keys that are suitable for biometrically authenticated secure communication.

How Reliable are Your Visual Attributes

Abstract: Describable visual attributes are a powerful way to label aspects of an image, and taken together, build a detailed representation of a scene’s appearance. Attributes enable highly accurate approaches to a variety of tasks, including object recognition, face recognition and image retrieval. An important consideration not previously addressed in the literature is the reliability of attribute classiers as the quality of an image degrades. In this paper, we introduce a general framework for conducting reliability studies that assesses attribute classier accuracy as a function of image degradation. This framework allows us to bound, in a probabilistic manner, the input imagery that is deemed acceptable for consideration by the attribute system { without requiring ground truth attribute labels. We introduce a novel dierential probabilistic model for accuracy assessment that leverages a strong normalization procedure based on the statistical extreme value theory. To demonstrate the utility of our framework, we present an extensive case study using 64 unique facial attributes, computed on data derived from the Labeled Faces in the Wild (LFW) data set. We also show that such reliability studies can result in signicant compression benets for mobile applications.

GRAB: generalized region assigned to binary

Abstract: Scale is one of the major challenges in recognition problems. For example, a face captured across large distances is considerably harder to recognize than the same face at small distances. Local binary pattern (LBP) and its variants have been successfully used in face detection, recognition, and many other computer vision applications. While LBP features are shown to be discriminative in face recognition, the pixel level description of LBP features is sensitive to the change in scale of the images. In this work, we extend the utility of a generalized variant of LBP feature descriptor called generalized region assigned to binary (GRAB), previously introduced in an article below, and show that it handles the challenges due to scale. The original LBP operator in another article is defined with respect to the surrounding pixel values while the GRAB operator is defined with respect to overlapping surrounding regions. This gives more general description and flexibility in choosing the right operator depending on the varying imaging conditions such as scale variations. We also propose a way to automatically select the scale of the GRAB operator (size of neighborhood). A pyramid of multi-scale GRAB operators is constructed, and the operator at each scale is applied to an image. Selection of operator’s scale is performed based on the number of stable pixels at different levels of the multi-scale pyramid. The stable pixels are defined to be the pixels in the images for which the GRAB value remains the same even as the GRAB operator scale changes. In addition to the experiments in the former article, we apply basic LBP, Liao et al.’s multi-scale block (MB)-LBP, and GRAB operator on face recognition across multiple scales and demonstrate that GRAB significantly outperforms the basic LBP and is more stable compared to MB-LBP in cases of reduced scale on a subsets of a well-known published database of labeled faces in the wild (LFW). We also perform experiments on the standard LFW database using strict LFW protocol and show the improved performance of GRAB descriptor compared to LBP and Gabor descriptors.

Trusted BWI: Privacy and trust enhanced biometric web identities

Abstract: Trusted web identities, which strongly associate a person with a digital identifier or certificate, are an area where biometrics should play a critical role. Balancing usability, security, and privacy is an important issue for any system that captures/stores users' information, especially for any biometric-based technology. To support biometric web services, the Biometric Identity Assurance Services (BIAS) standard was developed and recently approved. BIAS aims to establish standard biometric web services in order to improve interoperability and platform independence. Because they involve biometric data, the deployment of BIAS (and biometric web services in general) faces many challenges in terms of privacy, trust and security. They also face compatibility issues with widely-deployed systems that combine biometric sensors and Trusted Platform Modules (TPM). In order to address these obstacles, we propose an enhanced design of the recently introduced Biocryptographic Key Infrastructure (BKI). The original BKI enhanced the privacy and trust of remote biometric transactions, but, like most existing biometric systems, ignores the trust issues associated with remote enrollment. Our enhanced BKI design addresses this problem of trusted remote biometric enrollment. In addition, the enhanced design also extends the BKI to support biometric sensors with cryptographically secured on-chip biometric matching. Leveraging the new enhanced version of BKI, we propose the Trusted Biometric Web Identities (Trusted-BWI), as privacy and trust-enhanced biometric web services.

Pruning Bayesian Networks for Efficient Computation

Abstract: This paper analyzes the circumstances under which Bayesian networks can be pruned in order to reduce computational complexity without altering the computation for variables of interest. Given a problem instance which consists of a query and evidence for a set of nodes in the network, it is possible to delete portions of the network which do not participate in the computation for the query. Savings in computational complexity can be large when the original network is not singly connected. Results analogous to those described in this paper have been derived before [Geiger, Verma, and Pearl 89, Shachter 88] but the implications for reducing complexity of the computations in Bayesian networks have not been stated explicitly. We show how a preprocessing step can be used to prune a Bayesian network prior to using standard algorithms to solve a given problem instance. We also show how our results can be used in a parallel distributed implementation in order to achieve greater savings. We define a computationally equivalent subgraph of a Bayesian network. The algorithm developed in [Geiger, Verma, and Pearl 89] is modified to construct the subgraphs described in this paper with O(e) complexity, where e is the number of edges in the Bayesian network. Finally, we define a minimal computationally equivalent subgraph and prove that the subgraphs described are minimal.

Appearance, context and co-occurrence ensembles for identity recognition in personal photo collections

Abstract: While modern research in face recognition has focused on new feature representations, alternate learning methods for fusion of features, most have ignored the issue of unmodeled correlations in face data when combining diverse features such as similar visual regions, attributes, appearance frequency, etc. Conventional wisdom is that by using sufficient data and machine, one can learn the systematic correlations and use the data to form a more robust basis for core recognition tasks like verification, identification, and clustering. This however, takes large amounts of training data which is not really available for personal consumer photo collections. We address the fusion/correlation issue differently by proposing an ensemble-based approach that is built on different information sources such as facial appearance, visual context, and social (or co-occurrence) information of samples in a dataset, to provide higher classification accuracy for face recognition in consumer photo collections. To evaluate the utility of our ensembles and simultaneously generate stronger generic features, we perform two experiments - (i) a verification experiment on the standard unconstrained LFW (Labeled Faces in the Wild) dataset where by using an ensemble of appearance related features we report comparable results with recently reported state-of-the-art results and 2.9% better classification accuracy than the previous best method, and(ii) experiment on the Gallagher personal photo collection where we demonstrate at least 17% relative performance gain using visual context and social co-occurrence ensembles.

Issues in Rotational (Non-)invariance and Image Preprocessing

Abstract: This paper addresses two problems that have been largely overlooked in the literature. First, many systems seek to use, and algorithms claim to provide, rotational in-variance, such as fingerprint minutiae or SIFT/SURF features. We introduce a statistical test for rotational independence, using lossless rotations to show the differences are statistically significant and cannot be attributed to image noise. We use this to experimentally show fingerprint feature extractors fail to be rotation independent. We show the popular "rotation invariant" SURF and SIFT feature extractors, used in both biometric and general vision, also fail the rotation independence test. We then introduce a match-twist-match (MTM) paradigm and experimentally demonstrate that, by reducing the effective angular difference between probe and gallery, we can improve system matching performance. Our analysis, using FVC2002 and FVC2004 datasets, further shows that differences in extracted features impact the overall system performance of fingerprint matching of both matchers tested. Using the MTM approach, we reduce our secure template system's errors by 10%-20% -- helping us to define the current state of the art in the FVC-OnGoing Secure template competition with an EER of 1.698%. We end by bringing to the forefront the growing danger of sensors over-preprocessing of images. We show examples of the problems that can arise with preprocessing. As our rotation experiments showed, the impact of even modest numbers of feature errors suggest these preprocessing issues are likely very significant. We suggest the need for policy guidelines that require disclosure of preprocessing steps used and the development of standards for testing the impact of preprocessing.