+publisher:"University of North Carolina" +contributor:("Berg, Tamara")

University of North Carolina

1. Stanley, Natalie. Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks.

Degree: 2018, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:887b1903-bf0f-4667-b678-381ce5646a38

► Networks have become a common data mining tool to encode relational definitions between a set of entities. Whether studying biological correlations, or communication between individuals… (more)

Subjects/Keywords: School of Medicine; Curriculum in Bioinformatics and Computational Biology

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Stanley, N. (2018). Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:887b1903-bf0f-4667-b678-381ce5646a38

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Stanley, Natalie. “Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks.” 2018. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:887b1903-bf0f-4667-b678-381ce5646a38.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Stanley, Natalie. “Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks.” 2018. Web. 16 Jan 2021.

Vancouver:

Stanley N. Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks. [Internet] [Thesis]. University of North Carolina; 2018. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:887b1903-bf0f-4667-b678-381ce5646a38.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Stanley N. Adapting Community Detection Approaches to Large, Multilayer, and Attributed Networks. [Thesis]. University of North Carolina; 2018. Available from: https://cdr.lib.unc.edu/record/uuid:887b1903-bf0f-4667-b678-381ce5646a38

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

2. YANG, SHAN. NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS.

Degree: Computer Science, 2018, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:7841341b-857a-4e7d-9872-05d23b63e56d

► Material property has great importance in surgical simulation and virtual reality. The mechanical properties of the human soft tissue are critical to characterize the tissue… (more)

▼ Material property has great importance in surgical simulation and virtual reality. The mechanical properties of the human soft tissue are critical to characterize the tissue deformation of each patient. Studies have shown that the tissue stiffness described by the tissue properties may indicate abnormal pathological process. The (recovered) elasticity parameters can assist surgeons to perform better pre-op surgical planning and enable medical robots to carry out personalized surgical procedures. Traditional elasticity parameters estimation methods rely largely on known external forces measured by special devices and strain field estimated by landmarks on the deformable bodies. Or they are limited to mechanical property estimation for quasi-static deformation. For virtual reality applications such as virtual try-on, garment material capturing is of equal significance as the geometry reconstruction. In this thesis, I present novel approaches for automatically estimating the material properties of soft bodies from images or from a video capturing the motion of the deformable body. I use a coupled simulation-optimization-identification framework to deform one soft body at its original, non-deformed state to match the deformed geometry of the same object in its deformed state. The optimal set of material parameters is thereby determined by minimizing the error metric function. This method can simultaneously recover the elasticity parameters of multiple regions of soft bodies using Finite Element Method-based simulation (of either linear or nonlinear materials undergoing large deformation) and particle-swarm optimization methods. I demonstrate the effectiveness of this approach on real-time interaction with virtual organs in patient-specific surgical simulation, using parameters acquired from low-resolution medical images. With the recovered elasticity parameters and the age of the prostate cancer patients as features, I build a cancer grading and staging classifier. The classifier achieves up to 91% for predicting cancer T-Stage and 88% for predicting Gleason score. To recover the mechanical properties of soft bodies from a video, I propose a method which couples statistical graphical model with FEM simulation. Using this method, I can recover the material properties of a soft ball from a high-speed camera video that captures the motion of the ball. Furthermore, I extend the material recovery framework to fabric material identification. I propose a novel method for garment material extraction from a single-view image and a learning based cloth material recovery method from a video recording the motion of the cloth. Most recent garment capturing techniques rely on acquiring multiple views of clothing, which may not always be readily available, especially in the case of pre-existing photographs from the web. As an alternative, I propose a method that can compute a 3D model of a human body and its outfit from a single photograph with little human interaction. My proposed learning-based cloth material type… Advisors/Committee Members: YANG, SHAN, Lin, Ming, Berg, Tamara, Bregler, Chris, Manocha, Dinesh, Jojic, Vladimir, University of North Carolina at Chapel Hill.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

YANG, S. (2018). NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:7841341b-857a-4e7d-9872-05d23b63e56d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

YANG, SHAN. “NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS.” 2018. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:7841341b-857a-4e7d-9872-05d23b63e56d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

YANG, SHAN. “NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS.” 2018. Web. 16 Jan 2021.

Vancouver:

YANG S. NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS. [Internet] [Thesis]. University of North Carolina; 2018. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:7841341b-857a-4e7d-9872-05d23b63e56d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

YANG S. NON-RIGID BODY MECHANICAL PROPERTY RECOVERY FROM IMAGES AND VIDEOS. [Thesis]. University of North Carolina; 2018. Available from: https://cdr.lib.unc.edu/record/uuid:7841341b-857a-4e7d-9872-05d23b63e56d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

3. Ji, Dinghuang. Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements.

Degree: Computer Science, 2018, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:eaeceb64-ad74-416d-a0b6-b4ee48512f8d

► Our world is filled with living beings and other dynamic elements. It is important to record dynamic things and events for the sake of education,… (more)

▼ Our world is filled with living beings and other dynamic elements. It is important to record dynamic things and events for the sake of education, archeology, and culture inheritance. From vintage to modern times, people have recorded dynamic scene elements in different ways, from sequences of cave paintings to frames of motion pictures. This thesis focuses on two key computer vision techniques by which dynamic element representation moves beyond video capture: towards 3D reconstruction and view synthesis. Although previous methods on these two aspects have been adopted to model and represent static scene elements, dynamic scene elements present unique and difficult challenges for the tasks. This thesis focuses on three types of dynamic scene elements, namely 1) dynamic texture with static shape, 2) dynamic shapes with static texture, and 3) dynamic illumination of static scenes. Two research aspects will be explored to represent and visualize them: dynamic 3D reconstruction and dynamic view synthesis. Dynamic 3D reconstruction aims to recover the 3D geometry of dynamic objects and, by modeling the objects’ movements, bring 3D reconstructions to life. Dynamic view synthesis, on the other hand, summarizes or predicts the dynamic appearance change of dynamic objects – for example, the daytime-to-nighttime illumination of a building or the future movements of a rigid body. We first target the problem of reconstructing dynamic textures of objects that have (approximately) fixed 3D shape but time-varying appearance. Examples of such objects include waterfalls, fountains, and electronic billboards. Since the appearance of dynamic-textured objects can be random and complicated, estimating the 3D geometry of these objects from 2D images/video requires novel tools beyond the appearance-based point correspondence methods of traditional 3D computer vision. To perform this 3D reconstruction, we introduce a method that simultaneously 1) segments dynamically textured scene objects in the input images and 2) reconstructs the 3D geometry of the entire scene, assuming a static 3D shape for the dynamically textured objects. Compared to dynamic textures, the appearance change of dynamic shapes is due to physically defined motions like rigid body movements. In these cases, assumptions can be made about the object’s motion constraints in order to identify corresponding points on the object at different timepoints. For example, two points on a rigid object have constant distance between them in the 3D space, no matter how the object moves. Based on this assumption of local rigidity, we propose a robust method to correctly identify point correspondences of two images viewing the same moving object from different viewpoints and at different times. Dense 3D geometry could be obtained from the computed point correspondences. We apply this method on unsynchronized video streams, and observe that the number of inlier correspondences found by this method can be used as indicator for frame alignment among the different streams. … Advisors/Committee Members: Ji, Dinghuang, Frahm, Jan-Michael, Dunn, Enrique, Berg, Tamara, Niethammer, Marc, Savarese, Silvio.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Ji, D. (2018). Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:eaeceb64-ad74-416d-a0b6-b4ee48512f8d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Ji, Dinghuang. “Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements.” 2018. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:eaeceb64-ad74-416d-a0b6-b4ee48512f8d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Ji, Dinghuang. “Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements.” 2018. Web. 16 Jan 2021.

Vancouver:

Ji D. Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements. [Internet] [Thesis]. University of North Carolina; 2018. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:eaeceb64-ad74-416d-a0b6-b4ee48512f8d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Ji D. Data-driven 3D Reconstruction and View Synthesis of Dynamic Scene Elements. [Thesis]. University of North Carolina; 2018. Available from: https://cdr.lib.unc.edu/record/uuid:eaeceb64-ad74-416d-a0b6-b4ee48512f8d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

4. Liu, Wei. Localizing Objects Fast and Accurately.

Degree: Computer Science, 2016, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:46218f57-e071-4cf8-9f51-0065bede73f9

► A fundamental problem in computer vision is knowing what is in the image and where it is. We develop models to localize objects of multiple… (more)

▼ A fundamental problem in computer vision is knowing what is in the image and where it is. We develop models to localize objects of multiple categories, such as person and car, fast and accurately. In particular, we focus on designing deep convolutional neural networks (CNNs) for object detection and semantic segmentation. A central theme of this dissertation is to explore the design choices of network structure to combine the full power of CNNs and the characteristics of each task to not only achieve high-quality results but also keep the model relatively simple and fast. At the heart of object detection is the question of how to search efficiently through a continuous 2D bounding boxes space of various scales and aspect ratios at every possible location in an image. A brute force approach would be searching over all possibilities, but it is apparently not scalable and is quite difficult. An alternative is to propose some potential locations which might contain objects, and then classify each of the proposal. Because the search space is much smaller after the proposal step, we can use a more powerful feature to describe each proposal. A first contribution of this dissertation is to show that fine-tuning a much deeper network can boost the detection performance significantly, compared to a relatively shallower network. A second contribution of this dissertation is that we show that the search can be approximated by discretizing the search space and then regressing the residual difference between a discrete box and a target box. This is a departure from the proposal and then classify series of methods. We present a single stage framework, SSD, which can simultaneously detect and classify objects fast and accurately. SSD splits the space of small boxes more densely and the space of larger boxes more sparsely. As a result, it can discretize the space more efficiently and ease training notably. We have empirically shown that it is as accurate as or even better than the two-stage methods and yet is much faster. Unlike object detection, semantic segmentation is usually treated as a per-pixel classification problem, especially in the era of deep networks. However, a major issue is how to incorporate global semantic context information when making local decision. Although there are concurrent works on using techniques from graphical models such as conditional random fields (CRFs) to introduce context and structure information, we present a simple yet effective method, ParseNet, by using the average feature for a layer to augment the features at each location. Experimental results show that it can be as effective as a method which uses CRFs as a post-processing step to include context information. In order to make the above methods useful for many real-time systems, such as mobile devices or self-driving cars, we have collected large-scale video datasets for multiple categories, and hope that temporal consistency information in video can help further boost the performance and speed up the operations while lowering power… Advisors/Committee Members: Liu, Wei, Berg, Alexander, Berg, Tamara, Anguelov, Dragomir, Frahm, Jan-Michael, Niethammer, Marc.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Liu, W. (2016). Localizing Objects Fast and Accurately. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:46218f57-e071-4cf8-9f51-0065bede73f9

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Liu, Wei. “Localizing Objects Fast and Accurately.” 2016. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:46218f57-e071-4cf8-9f51-0065bede73f9.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Liu, Wei. “Localizing Objects Fast and Accurately.” 2016. Web. 16 Jan 2021.

Vancouver:

Liu W. Localizing Objects Fast and Accurately. [Internet] [Thesis]. University of North Carolina; 2016. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:46218f57-e071-4cf8-9f51-0065bede73f9.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Liu W. Localizing Objects Fast and Accurately. [Thesis]. University of North Carolina; 2016. Available from: https://cdr.lib.unc.edu/record/uuid:46218f57-e071-4cf8-9f51-0065bede73f9

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

5. Vittayakorn, Sirion. Visual attribute discovery and analyses from Web data.

Degree: Computer Science, 2016, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:e99f13f0-8689-49cc-a15d-3269cb0e4732

► Visual attributes are important for describing and understanding an object’s appearance. For an object classification or recognition task, an algorithm needs to infer the visual… (more)

▼ Visual attributes are important for describing and understanding an object’s appearance. For an object classification or recognition task, an algorithm needs to infer the visual attributes of an object to compare, categorize or recognize the objects. In a zero-shot learning scenario, the algorithm depends on the visual attributes to describe an unknown object since the training samples are not available. Because different object categories usually share some common attributes (e.g. many animals have four legs, a tail and fur), the act of explicitly modeling attributes not only allows previously learnt attributes to be transferred to a novel category but also reduces the number of training samples for the new category which can be important when the number of training samples is limited. Even though larger numbers of visual attributes help the algorithm to better describe an image, they also require a larger set of training data. In the supervised scenario, data collection can be both a costly and time-consuming process. To mitigate the data collection costs, this dissertation exploits the weakly-supervised data from the Web in order to construct computational methodologies for the discovery of visual attributes, as well as an analysis across time and domains. This dissertation first presents an automatic approach to learning hundreds of visual attributes from the open-world vocabulary on the Web using a convolutional neural network. The proposed method tries to understand visual attributes in terms of perception inside deep neural networks. By focusing on the analysis of neural activations, the system can identify the degree to which an attribute can be visually perceptible and can localize the visual attributes in an image. Moreover, the approach exploits the layered structure of the deep model to determine the semantic depth of the attributes. Beyond visual attribute discovery, this dissertation explores how visual styles (i.e., attributes that correspond to multiple visual concepts) change across time. These are referred to as visual trends. To this goal, this dissertation introduces several deep neural networks for estimating when objects were made together with the analyses of the neural activations and their degree of entropy to gain insights into the deep network. To utilize the dating of the historical object frameworks in real-world applications, the dating frameworks are applied to analyze the influence of vintage fashion on runway collections, as well as to analyze the influence of fashion on runway collections and on street fashion. Finally, this dissertation introduces an approach to recognizing and transferring visual attributes across domains in a realistic manner. Given two input images from two different domains: 1) a shopping image, and 2) a scene image, this dissertation proposes a generative adversarial network for transferring the product pixels from the shopping image to the scene image such that: 1) the output image looks realistic and 2) the visual attributes of the product are preserved. In… Advisors/Committee Members: Vittayakorn, Sirion, Berg, Tamara, Berg, Alexander, Frahm, Jan-Michael, Hays, James, Fidler, Sanja.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Vittayakorn, S. (2016). Visual attribute discovery and analyses from Web data. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:e99f13f0-8689-49cc-a15d-3269cb0e4732

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Vittayakorn, Sirion. “Visual attribute discovery and analyses from Web data.” 2016. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:e99f13f0-8689-49cc-a15d-3269cb0e4732.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Vittayakorn, Sirion. “Visual attribute discovery and analyses from Web data.” 2016. Web. 16 Jan 2021.

Vancouver:

Vittayakorn S. Visual attribute discovery and analyses from Web data. [Internet] [Thesis]. University of North Carolina; 2016. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:e99f13f0-8689-49cc-a15d-3269cb0e4732.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Vittayakorn S. Visual attribute discovery and analyses from Web data. [Thesis]. University of North Carolina; 2016. Available from: https://cdr.lib.unc.edu/record/uuid:e99f13f0-8689-49cc-a15d-3269cb0e4732

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

6. Xu, Yi. Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints.

Degree: Computer Science, 2016, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:926a5d94-944d-4c61-963f-b863b0dc1f41

► The continuous stream of videos that are uploaded and shared on the Internet has been leveraged by computer vision researchers for a myriad of detection… (more)

▼ The continuous stream of videos that are uploaded and shared on the Internet has been leveraged by computer vision researchers for a myriad of detection and retrieval tasks, including gesture detection, copy detection, face authentication, etc. However, the existing state-of-the-art event detection and retrieval techniques fail to deal with several real-world challenges (e.g., low resolution, low brightness and noise) under adversary constraints. This dissertation focuses on these challenges in realistic scenarios and demonstrates practical methods to address the problem of robustness and efficiency within video event detection and retrieval systems in five application settings (namely, CAPTCHA decoding, face liveness detection, reconstructing typed input on mobile devices, video confirmation attack, and content-based copy detection). Specifically, for CAPTCHA decoding, I propose an automated approach which can decode moving-image object recognition (MIOR) CAPTCHAs faster than humans. I showed that not only are there inherent weaknesses in current MIOR CAPTCHA designs, but that several obvious countermeasures (e.g., extending the length of the codeword) are not viable. More importantly, my work highlights the fact that the choice of underlying hard problem selected by the designers of a leading commercial solution falls into a solvable subclass of computer vision problems. For face liveness detection, I introduce a novel approach to bypass modern face authentication systems. More specifically, by leveraging a handful of pictures of the target user taken from social media, I show how to create realistic, textured, 3D facial models that undermine the security of widely used face authentication solutions. My framework makes use of virtual reality (VR) systems, incorporating along the way the ability to perform animations (e.g., raising an eyebrow or smiling) of the facial model, in order to trick liveness detectors into believing that the 3D model is a real human face. I demonstrate that such VR-based spoofing attacks constitute a fundamentally new class of attacks that point to a serious weaknesses in camera-based authentication systems. For reconstructing typed input on mobile devices, I proposed a method that successfully transcribes the text typed on a keyboard by exploiting video of the user typing, even from significant distances and from repeated reflections. This feat allows us to reconstruct typed input from the image of a mobile phone’s screen on a user’s eyeball as reflected through a nearby mirror, extending the privacy threat to include situations where the adversary is located around a corner from the user. To assess the viability of a video confirmation attack, I explored a technique that exploits the emanations of changes in light to reveal the programs being watched. I leverage the key insight that the observable emanations of a display (e.g., a TV or monitor) during presentation of the viewing content induces a distinctive flicker pattern that can be exploited by an adversary. My proposed approach… Advisors/Committee Members: Xu, Yi, Frahm, Jan-Michael, Monrose, Fabian, Dunn, Enrique, Berg, Tamara, Crandall, David.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Xu, Y. (2016). Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:926a5d94-944d-4c61-963f-b863b0dc1f41

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Xu, Yi. “Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints.” 2016. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:926a5d94-944d-4c61-963f-b863b0dc1f41.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Xu, Yi. “Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints.” 2016. Web. 16 Jan 2021.

Vancouver:

Xu Y. Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints. [Internet] [Thesis]. University of North Carolina; 2016. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:926a5d94-944d-4c61-963f-b863b0dc1f41.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Xu Y. Toward Robust Video Event Detection and Retrieval Under Adversarial Constraints. [Thesis]. University of North Carolina; 2016. Available from: https://cdr.lib.unc.edu/record/uuid:926a5d94-944d-4c61-963f-b863b0dc1f41

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

7. Han, Xufeng. LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION.

Degree: Computer Science, 2016, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:e712c6a4-e4ff-4c3b-8646-a6c1c872f042

► This thesis studies machine learning problems involved in visual recognition on a variety of computer vision tasks. It attacks the challenge of scaling-up learning to… (more)

▼ This thesis studies machine learning problems involved in visual recognition on a variety of computer vision tasks. It attacks the challenge of scaling-up learning to efficiently handle more training data in object recognition, more noise in brain activation patterns, and learning more capable visual similarity models. For learning similarity models, one challenge is to capture from data the subtle correlations that preserve the notion of similarity relevant to the task. Most previous work focused on improving feature learning and metric learning separately. Instead, we propose a unified deep-learning modeling framework that jointly optimizes the two through back-propagation. We model the feature mapping using a convolutional neural network and the metric function using a multi-layer fully-connected network. Enabled by large datasets and a sampler to handle the intrinsic imbalance between positive and negative samples, we are able to learn such models efficiently. We apply this approach to patch-based image matching and cross-domain clothing-item matching. For analyzing activation patterns in images acquired using functional Magnetic Resonance Imaging (fMRI), a technology widely used in neuroscience to study human brain, challenges are small number of examples and high level of noise. The common ways of increasing the signal to noise ratio include adding more repetitions, averaging trials, and analyzing statistics maps solved based on a general linear model. In collaboration with neuroscientists, we developed a machine learning approach that allows us to analyze individual trials directly. This approach uses multi-voxel patterns over regions of interest as feature representation, and helps discover effects previous analyses missed. For multi-class object recognition, one challenge is learning a non-one-vs-all multi-class classifier with large numbers of categories each with large numbers of examples. A common approach is data parallelization in a synchronized fashion: evenly and randomly distribute the data into splits, learn a full model on each split and average the models. We reformulate the overall learning problem in a consensus optimization framework and propose a more principled synchronized approach to distributed training. Moreover, we develop an efficient algorithm for solving the sub-problem by reducing it to a standard problem with warm start. Advisors/Committee Members: Han, Xufeng, Berg, Alexander, Berg, Alexander, Berg, Tamara, Leung, Thomas, McMillan, Leonard, Niethammer, Marc.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Han, X. (2016). LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:e712c6a4-e4ff-4c3b-8646-a6c1c872f042

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Han, Xufeng. “LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION.” 2016. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:e712c6a4-e4ff-4c3b-8646-a6c1c872f042.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Han, Xufeng. “LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION.” 2016. Web. 16 Jan 2021.

Vancouver:

Han X. LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION. [Internet] [Thesis]. University of North Carolina; 2016. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:e712c6a4-e4ff-4c3b-8646-a6c1c872f042.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Han X. LEARNING WITH MORE DATA AND BETTER MODELS FOR VISUAL SIMILARITY AND DIFFERENTIATION. [Thesis]. University of North Carolina; 2016. Available from: https://cdr.lib.unc.edu/record/uuid:e712c6a4-e4ff-4c3b-8646-a6c1c872f042

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

8. Zheng, Enliang. TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS.

Degree: Computer Science, 2016, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:adaabbef-0021-4d4b-87b3-4b6f9985d97d

► The goal of image-based 3D reconstruction is to construct a spatial understanding of the world from a collection of images. For applications that seek to… (more)

▼ The goal of image-based 3D reconstruction is to construct a spatial understanding of the world from a collection of images. For applications that seek to model generic real-world scenes, it is important that the reconstruction methods used are able to characterize both static scene elements (e.g. trees and buildings) as well as dynamic objects (e.g. cars and pedestrians). However, due to many inherent ambiguities in the reconstruction problem, recovering this 3D information with accuracy, robustness, and efficiency is a considerable challenge. To advance the research frontier for image-based 3D modeling, this dissertation focuses on three challenging problems in static scene and dynamic object reconstruction. We first target the problem of static scene depthmap estimation from crowd-sourced datasets (i.e. photos collected from the Internet). While achieving high-quality depthmaps using images taken under a controlled environment is already a difficult task, heterogeneous crowd-sourced data presents a unique set of challenges for multi-view depth estimation, including varying illumination and occasional occlusions. We propose a depthmap estimation method that demonstrates high accuracy, robustness, and scalability on a large number of photos collected from the Internet. Compared to static scene reconstruction, the problem of dynamic object reconstruction from monocular images is fundamentally ambiguous when not imposing any additional assumptions. This is because having only a single observation of an object is insufficient for valid 3D triangulation, which typically requires concurrent observations of the object from multiple viewpoints. Assuming that dynamic objects of the same class (e.g. all the pedestrians walking on a sidewalk) move in a common path in the real world, we develop a method that estimates the 3D positions of the dynamic objects from unstructured monocular images. Experiments on both synthetic and real datasets illustrate the solvability of the problem and the effectiveness of our approach. Finally, we address the problem of dynamic object reconstruction from a set of unsynchronized videos capturing the same dynamic event. This problem is of great interest because, due to the increased availability of portable capture devices, captures using multiple unsynchronized videos are common in the real world. To resolve the challenges that arises from non-concurrent captures and unknown temporal overlap among video streams, we propose a self-expressive dictionary learning framework, where the dictionary entries are defined as the collection of temporally varying structures. Experiments demonstrate the effectiveness of this approach to the previously unsolved problem. Advisors/Committee Members: Zheng, Enliang, Frahm, Jan-Michael, Dunn, Enrique, Berg, Tamara, Jojic, Vladimir, Sheikh, Yaser.

Subjects/Keywords: College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Zheng, E. (2016). TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:adaabbef-0021-4d4b-87b3-4b6f9985d97d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Zheng, Enliang. “TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS.” 2016. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:adaabbef-0021-4d4b-87b3-4b6f9985d97d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Zheng, Enliang. “TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS.” 2016. Web. 16 Jan 2021.

Vancouver:

Zheng E. TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS. [Internet] [Thesis]. University of North Carolina; 2016. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:adaabbef-0021-4d4b-87b3-4b6f9985d97d.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Zheng E. TOWARD 3D RECONSTRUCTION OF STATIC AND DYNAMIC OBJECTS. [Thesis]. University of North Carolina; 2016. Available from: https://cdr.lib.unc.edu/record/uuid:adaabbef-0021-4d4b-87b3-4b6f9985d97d

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

9. Kiapour, Mohammadhadi. LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES.

Degree: Computer Science, 2015, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:d5241918-b3f4-4089-86de-f9d957179775

► Clothing recognition is a societally and commercially important yet extremely challenging problem due to large variations in clothing appearance, layering, style, body shape and pose.… (more)

▼ Clothing recognition is a societally and commercially important yet extremely challenging problem due to large variations in clothing appearance, layering, style, body shape and pose. In this dissertation, we propose new computational vision approaches that learn to represent and recognize clothing items in images. First, we present an effective method for parsing clothing in fashion photographs, where we label the regions of an image with their clothing categories. We then extend our approach to tackle the clothing parsing problem using a data-driven methodology: for a query image, we find similar styles from a large database of tagged fashion images and use these examples to recognize clothing items in the query. Along with our novel large fashion dataset, we also present intriguing initial results on using clothing estimates to improve human pose identification. Second, we examine questions related to fashion styles and identifying the clothing elements associated with each style. We first design an online competitive style rating game called Hipster Wars to crowd source reliable human judgments of clothing styles. We use this game to collect a new dataset of clothing outfits with associated style ratings for different clothing styles. Next, we build visual style descriptors and train models that are able to classify clothing styles and identify the clothing elements are most discriminative in every style. Finally, we define a new task, Exact Street to Shop, where our goal is to match a real-world example of a garment item to the same exact garment in an online shop. This is an extremely challenging task due to visual differences between street photos that are taken of people wearing clothing in everyday uncontrolled settings, and online shop photos, which are captured by professionals in highly controlled settings. We introduce a novel large dataset for this application, collected from the web, and present a deep learning based similarity network that can compare clothing items across visual domains. Advisors/Committee Members: Kiapour, Mohammadhadi, Berg, Tamara, Berg, Tamara, Berg, Alexander, Lazebnik, Svetlana, Frahm, Jan-Michael, Piramuthu, Robinson.

Subjects/Keywords: Computer science; College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Kiapour, M. (2015). LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:d5241918-b3f4-4089-86de-f9d957179775

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Kiapour, Mohammadhadi. “LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES.” 2015. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:d5241918-b3f4-4089-86de-f9d957179775.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Kiapour, Mohammadhadi. “LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES.” 2015. Web. 16 Jan 2021.

Vancouver:

Kiapour M. LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES. [Internet] [Thesis]. University of North Carolina; 2015. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:d5241918-b3f4-4089-86de-f9d957179775.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Kiapour M. LARGE SCALE VISUAL RECOGNITION OF CLOTHING, PEOPLE AND STYLES. [Thesis]. University of North Carolina; 2015. Available from: https://cdr.lib.unc.edu/record/uuid:d5241918-b3f4-4089-86de-f9d957179775

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

University of North Carolina

10. Ordonez Roman, Vicente. Language and Perceptual Categorization in Computational Visual Recognition.

Degree: Computer Science, 2015, University of North Carolina

URL: https://cdr.lib.unc.edu/record/uuid:188ef51f-d3dc-4216-97ea-07da5109a1a6

► Computational visual recognition or giving computers the ability to understand images as well as humans do is a core problem in Computer Vision. Traditional recognition… (more)

▼ Computational visual recognition or giving computers the ability to understand images as well as humans do is a core problem in Computer Vision. Traditional recognition systems often describe visual content by producing a set of isolated labels, object locations, or by even trying to annotate every pixel in an image with a category. People instead describe the visual world using language. The rich visually descriptive language produced by people incorporates information from human intuition, world knowledge, visual saliency, and common sense that go beyond detecting individual visual concepts like objects, attributes, or scenes. Moreover, due to the rising popularity of social media, there exist billions of images with associated text on the web, yet systems that can leverage this type of annotations or try to connect language and vision are scarce. In this dissertation, we propose new approaches that explore the connections between language and vision at several levels of detail by combining techniques from Computer Vision and Natural Language Understanding. We first present a data-driven technique for understanding and generating image descriptions using natural language, including automatically collecting a big-scale dataset of images with visually descriptive captions. Then we introduce a system for retrieving short visually descriptive phrases for describing some part or aspect of an image, and a simple technique to generate full image descriptions by stitching short phrases. Next we introduce an approach for collecting and generating referring expressions for objects in natural scenes at a much larger scale than previous studies. Finally, we describe methods for learning how to name objects by using intuitions from perceptual categorization related to basic-level and entry-level categories. The main contribution of this thesis is in advancing our knowledge on how to leverage language and intuitions from human perception to create visual recognition systems that can better learn from and communicate with people. Advisors/Committee Members: Ordonez Roman, Vicente, Berg, Tamara, Berg, Alexander, Efros, Alexei, Choi, Yejin, Frahm, Jan-Michael.

Subjects/Keywords: Computer science; College of Arts and Sciences; Department of Computer Science

Record Details Similar Records Cite « Share

❌

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6^th Edition):

Ordonez Roman, V. (2015). Language and Perceptual Categorization in Computational Visual Recognition. (Thesis). University of North Carolina. Retrieved from https://cdr.lib.unc.edu/record/uuid:188ef51f-d3dc-4216-97ea-07da5109a1a6

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Chicago Manual of Style (16^th Edition):

Ordonez Roman, Vicente. “Language and Perceptual Categorization in Computational Visual Recognition.” 2015. Thesis, University of North Carolina. Accessed January 16, 2021. https://cdr.lib.unc.edu/record/uuid:188ef51f-d3dc-4216-97ea-07da5109a1a6.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

MLA Handbook (7^th Edition):

Ordonez Roman, Vicente. “Language and Perceptual Categorization in Computational Visual Recognition.” 2015. Web. 16 Jan 2021.

Vancouver:

Ordonez Roman V. Language and Perceptual Categorization in Computational Visual Recognition. [Internet] [Thesis]. University of North Carolina; 2015. [cited 2021 Jan 16]. Available from: https://cdr.lib.unc.edu/record/uuid:188ef51f-d3dc-4216-97ea-07da5109a1a6.

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

Council of Science Editors:

Ordonez Roman V. Language and Perceptual Categorization in Computational Visual Recognition. [Thesis]. University of North Carolina; 2015. Available from: https://cdr.lib.unc.edu/record/uuid:188ef51f-d3dc-4216-97ea-07da5109a1a6

Note: this citation may be lacking information needed for this citation format:
Not specified: Masters Thesis or Doctoral Dissertation

		in
And / Or
		in
And / Or
		in
And / Or
		in

Open Access Theses and Dissertations