Mohammad Nayeem Teli

The challenge of face recognition from digital point-and-shoot cameras

Inexpensive “point-and-shoot” camera technology has combined with social network technology to give the general population a motivation to use face recognition technology. Users expect a lot; they want to snap pictures, shoot videos, upload, and have their friends, family and acquaintances more-or-less automatically recognized. Despite the apparent simplicity of the problem, face recognition in this context is hard. Roughly speaking, failure rates in the 4 to 8 out of 10 range are common. In contrast, error rates drop to roughly 1 in 1,000 for well controlled imagery. To spur advancement in face and person recognition this paper introduces the Point-and-Shoot Face Recognition Challenge (PaSC). The challenge includes 9,376 still images of 293 people balanced with respect to distance to the camera, alternative sensors, frontal versus not-frontal views, and varying location. There are also 2,802 videos for 265 people: a subset of the 293. Verification results are presented for public baseline algorithms and a commercial algorithm for three cases: comparing still images to still images, videos to videos, and still images to videos.

When high-quality face images match poorly

In face recognition, quality is typically thought of as a property of individual images, not image pairs. The implicit assumption is that high-quality images should be easy to match to each other, while low quality images should be hard to match. This paper presents a relational graph-based evaluation technique that uses match scores produced by face recognition algorithms to determine the “quality” of images. The resulting analysis demonstrates that only a small fraction of the images in a well-studied data set (FRVT 2006) are low-quality images. It is much more common to find relationships in which two images that are hard to match to each other can be easily matched with other images of the same person. In other words, these images are simultaneously both high and low quality. The existence of such contrary images represents a fundamental challenge for approaches to biometric quality that cast quality as an intrinsic property of a single image. Instead it indicates that quality should be associated with pairs of images. In exploring these contrary images, we find a surprising dependence on whether elements of an image pair are acquired at the same location, even in circumstances where one would be tempted to think of the locations as interchangeable. The results presented have important implications for anyone designing face recognition evaluations as well as those developing new algorithms.

Biometric zoos: Theory and experimental evidence

Several studies have shown the existence of biometric zoos. The premise is that in biometric systems people fall into distinct categories, labeled with animal names, indicating recognition difficulty. Different combinations of excessive false accepts or rejects correspond to labels such as: Goat, Lamb, Wolf, etc. Previous work on biometric zoos has investigated the existence of zoos for the results of an algorithm on a data set. This work investigates biometric zoos generalization across algorithms and data sets. For example, if a subject is a Goat for algorithm A on data set X, is that subject also a Goat for algorithm B on data set Y? This paper introduces a theoretical framework for generalizing biometric zoos. Based on our framework, we develop an experimental methodology for determining if biometric zoos generalize across algorithms and data sets, and we conduct a series of experiments to investigate the existence of zoos on two algorithms in FRVT 2006.

On the existence of face quality measures

We investigate the existence of quality measures for face recognition. First, we introduce the concept of an oracle for image quality in the context of face recognition. Next we introduce greedy pruned ordering (GPO) as an approximation to an image quality oracle. GPO analysis provides an estimated upper bound for quality measures, given a face recognition algorithm and data set. We then assess the performance of 12 commonly proposed face image quality measures against this standard. In addition, we investigate the potential for learning new quality measures via supervised learning. Finally, we show that GPO analysis is applicable to other biometrics.

Face and eye detection on hard datasets

Face and eye detection algorithms are deployed in a wide variety of applications. Unfortunately, there has been no quantitative comparison of how these detectors perform under difficult circumstances. We created a dataset of low light and long distance images which possess some of the problems encountered by face and eye detectors solving real world problems. The dataset we created is composed of reimaged images (photohead) and semi-synthetic heads imaged under varying conditions of low light, atmospheric blur, and distances of 3m, 50m, 80m, and 200m. This paper analyzes the detection and localization performance of the participating face and eye algorithms compared with the Viola Jones detector and four leading commercial face detectors. Performance is characterized under the different conditions and parameterized by per-image brightness and contrast. In localization accuracy for eyes, the groups/companies focusing on long-range face detection outperform leading commercial applications.

Pose manifold curvature is typically less near frontal face views

This research presents a study of the geometry of the face manifold as a person changes their horizontal pose from one profile to another. Although, a lot of research has gone into aspects of determining an ideal pose for pose invariant face recognition, less has been done to present the manifold of the faces presented by these pose variations. The novelty of our approach lies in the presentation of a finely sampled profile-to-profile dataset that is analyzed using Locally Linear Embedding (LLE) to estimate the curvature of these manifolds. Our results indicate that the profile-to-profile manifold is less curved, hence more linear, in the region around the frontal view than for any other region of the manifold, i.e. pose.

Nonlinear dimensionality reduction of electroencephalogram (EEG) for Brain Computer interfaces

Patterns in electroencephalogram (EEG) signals are analyzed for a Brain Computer Interface (BCI). An important aspect of this analysis is the work on transformations of high dimensional EEG data to low dimensional spaces in which we can classify the data according to mental tasks being performed. In this research we investigate how a Neural Network (NN) in an auto-encoder with bottleneck configuration can find such a transformation. We implemented two approximate second-order methods to optimize the weights of these networks, because the more common first-order methods are very slow to converge for networks like these with more than three layers of computational units. The resulting non-linear projections of time embedded EEG signals show interesting separations that are related to tasks. The bottleneck networks do indeed discover nonlinear transformations to low-dimensional spaces that capture much of the information present in EEG signals. However, the resulting low-dimensional representations do not improve classification rates beyond what is possible using Quadratic Discriminant Analysis (QDA) on the original time-lagged EEG.