The sharing of data has become integral to advancing scientific research, but it introduces challenges related to safeguarding personally identifiable information (PII). This thesis addresses the specific problem of sharing drivers' face videos for transportation research while ensuring privacy protection. To tackle this issue, we leverage recent advancements in generative adversarial networks (GANs) and demonstrate their effectiveness in deidentifying individuals by swapping their faces with those of others. Extensive experimentation is conducted using a large-scale dataset from ORNL, enabling the quantification of errors associated with head movements, mouth movements, eye movements, and other human factors cues. Additionally, qualitative analysis using metrics such as PERCLOS (Percentage of Eye Closure) and human evaluators provide valuable insights into the quality and fidelity of the deidentified videos. To enhance privacy preservation, we propose the utilization of synthetic faces as substitutes for real faces. Moreover, we introduce practical guidelines, including the establishment of thresholds and spot checking, to incorporate human-in-the-loop validation, thereby improving the accuracy and reliability of the deidentification process. In addition to this, this thesis also presents mitigation strategies to effectively handle reidentification risks. By considering the potential exploitation of soft biometric identifiers or non-biometric cues, we highlight the importance of implementing comprehensive measures such as robust data user licenses and privacy protection protocols. / Master of Science / With the increasing availability of large-scale datasets in transportation engineering, ensuring the privacy and confidentiality of sensitive information has become a paramount concern. One specific area of concern is the protection of drivers' facial data captured by the National Driving Simulator (NDS) during research studies. The potential risks associated with the misuse or unauthorized access to such data necessitate the development of robust deidentification techniques. In this thesis, we propose a GAN-based framework for the deidentification of drivers' face videos while preserving important facial attribute information. The effectiveness of the proposed framework is evaluated through comprehensive experiments, considering various metrics related to human factors. The results demonstrate the capability of the framework to successfully deidentify face videos, enabling the safe sharing and analysis of valuable transportation research data. This research contributes to the field of transportation engineering by addressing the critical need for privacy protection while promoting data sharing and advancing human factors research.

The bitwist construction, originally discovered by Cannon, Floyd, and Parry, gives us a new method for finding face pairing descriptions of 3-manifolds. In this paper, I will describe the construction in a way suitable for a more general audience than the original research papers. Along the way, I will describe Dehn Surgery and a set of moves which allows us to change the framings of a link without changing the topology of the manifold obtained by Dehn Surgery. Once the theory has been developed, I will apply it to find several bitwist representations of the Poincaré Sphere and 3-Torus. Finally, I discuss how one might attempt to find a set of moves that can take one bitwist representation of a manifold to any other bitwist representation of the same manifold. / Master of Science

Real-time face recognition has recently become available for the government and industry due to developments in face recognition algorithms, human head detection algorithms, and faster/low cost computers. Despite these advances, however, there are still some critical issues that affect the performance of real-time face recognition software. This paper addresses the problem of off-centered and out-of-pose faces in pictures, particularly in regard to the eigenface method for face recognition. We first demonstrate how the representation of faces by the eigenface method, and ultimately the performance of the software depend on the location of the eyes in the pictures. The eigenface method for face recognition is described: specifically, the creation of a face basis using the singular value decomposition, the reduction of dimension, and the unique representation of faces in the basis. Two different approaches for aligning the eyes in images are presented. The first considers the rotation of images using the orthogonal Procrustes Problem. The second approach looks at locating features in images using energy-minimizing active contours. We then conclude with a simple and fast algorithm for locating faces in images. Future research is also discussed. / Master of Science

Engaging in productive political discussion has long been a valued aspect of American democratic life. Due to ease of access and the potential for exposure to diverse views, the Internet and social media may support mediated political talk. Literature on the concept of face and politeness theory provides a framework for understanding interpersonal interactions, both online and offline. To understand if social media has the potential to host political discussion among millennials, a survey (N = 352) of undergraduate students examined social media use and political interaction experiences. Facebook was the most popular platform for exposure to others' political opinions and political self-expression. Facebook users with more diverse networks engaged in more political expression. Across numerous platforms, participants reported frequently being exposed to others' political opinions but infrequently sharing their own views. Negative and positive political interactions on Facebook and Twitter were explored for their threat to and support of negative face (need for autonomy) and positive face (need for validation). Findings indicate that engaging in negative interactions leads to more face threat while observing negative interactions solicits more face support. Engaging in positive interactions results in more face support and observing positive interactions leads to more face threat. Across interaction type and platform, participants who actively engaged in political interactions as opposed to merely observing them reported significantly more subsequent online political engagement. Future research on political interactions across various social media platforms and the application of interpersonal communication theory to the study of mediated political talk is warranted. / Master of Arts

Preliminary experiments in Virginia by Wallace and Turner (1961) on face fly control indicated that there was some promise using low-level feeding of a chemical such as ronnel in a mineral salt mixture as a larvicide. In cooperation with the Moorman Manufacturing Company and Mr. H. C. Stuart of Elk Garden, Virginia, a large scale experiment was initiated in the spring of 1962 . Approximately 1600 head of cattle in 23 pastures were utilized in this experiment. All of the pastures received a mineral salt mixture containing 5.5 percent ronnel with the exception of two pastures on the perimeter of the treated area that received no ronnel and were used as checks. Adult and larval counts were made on 7 of the 23 treated pastures and both of the untreated pastures. Results indicated that: (1) In spite of good larval control adult fly counts remained high, especially in treated pastures near untreated areas. Apparently the face fly has a natural tendency for dispersion, and, therefore, area control by low-level feeding of salt containing a larvicide is difficult. (2) When the consumption of ronnel remained above 5.5 mg per kg of animal body weight per day, larval control was above 95 percent. (3) During hot, dry periods animals reduced their salt consumption and this resulted in decreased larval control. (4) When salt boxes were placed in areas where cattle were seldom seen to frequent, salt consumption was lowered, as was larval control. / Master of Science

In 2-dimensional topology, quotients of polygons by edge-pairings provide a rich source of examples of closed, connected, orientable surfaces. In fact, they provide all such examples. The 3-dimensional analogue of an edge-pairing of a polygon is a face-pairing of a faceted 3-ball. Unfortunately, quotients of faceted 3-balls by face-pairings rarely provide us with examples of 3-manifolds due to singularities that arise at the vertices. However, any face-pairing of a faceted 3-ball may be slighted modified so that its quotient is a genuine manifold, i.e. free of singularities. The modified face-pairing is called a twisted face-pairing. It is natural to ask which closed, connected, orientable 3-manifolds may be obtained as quotients of twisted face-pairings. In this paper, we focus on a special class of face-pairings called cycle-free twisted face-pairings and give description of their quotient spaces in terms of integer weighted graphs. We use this description to prove that most spherical 3-manifolds can be obtained as quotients of cycle-free twisted face-pairings, but the Poincaré homology 3-sphere cannot. / Master of Science

Naive observers are found to estimate numbers of dots on projected slides with a consistent factor of underestimation of ca. 0.75. There is an overall difference in response between males and females and between more experienced and less experienced observers. There is an apparent trade-off between accuracy and precision when significant differences in response are recorded. Training observers with slides of revealed value alone is equally effective as a combination of slides and a reference pamphlet: both produce improvements in accuracy and precision. A subjective visual index of house fly abundance is found to be more sensitive than Scudder grill counts, spot cards, sticky tape traps or square-foot panels to changes in house fly population densities. None of these methods is consistently the best correlated to the visual index, although the square-foot panels were most highly correlated most often. / Ph. D.

Yes / To introduce and evaluate the performance in detecting glaucomatous abnormalities of a novel method for extracting en face slab images (SMAS), which considers varying individual anatomy and configuration of retinal nerve fiber bundles. Dense central retinal spectral domain optical coherence tomography scans were acquired in 16 participants with glaucoma and 19 age-similar controls. Slab images were generated by averaging reflectivity over different depths below the inner limiting membrane according to several methods. SMAS considered multiple 16 µm thick slabs from 8 to 116 µm below the inner limiting membrane, whereas 5 alternative methods considered single summary slabs of various thicknesses and depths. Superpixels in eyes with glaucoma were considered abnormal if below the first percentile of distributions fitted to control data for each method. The ability to detect glaucoma defects was measured by the proportion of abnormal superpixels. Proportion of superpixels below the fitted first percentile in controls was used as a surrogate false-positive rate. The effects of slab methods on performance measures were evaluated with linear mixed models. The ability to detect glaucoma defects varied between slab methods, χ2(5) = 120.9, P

Yes / To assess how objectively-detected defects of retinal nerve fibre bundle (RNFB) reflectance in en face OCT images relate to circumpapillary retinal nerve fibre layer thickness (cpRNFLT) and visual field defects. Cross-sectional study. 16 participants with early-glaucoma and 29 age-similar healthy controls, among which 22 had usable en face images to establish normative levels of RNFB reflectance. All participants underwent cpRNFLT scans, visual field examination and wide-field OCT. En face reflectivity was assessed objectively using the Summary of Multiple Anatomically-Adjusted Slabs (SMAS) method. En face defects were deemed concordant with cpRNFLT when they had at least one cpRNFLT point with p

Yes / This paper discusses how a boundary-based surface fitting approach can be utilised to smoothly reconstruct a given human face where the scan data corresponding to the face is provided. In particular, the paper discusses how a solution to the Biharmonic equation can be used to set up the corresponding boundary value problem. We show how a compact explicit solution method can be utilised for efficiently solving the chosen Biharmonic equation. Thus, given the raw scan data of a 3D face, we extract a series of profile curves from the data which can then be utilised as boundary conditions to solve the Biharmonic equation. The resulting solution provides us a continuous single surface patch describing the original face.