Theses and Dissertations
This collection includes both ASU Theses and Dissertations, submitted by graduate students, and the Barrett, Honors College theses submitted by undergraduate students.
Displaying 1 - 10 of 26
Filtering by
- All Subjects: Computer Engineering
- All Subjects: deep learning
- Creators: Turaga, Pavan
Description
Motion capture using cost-effective sensing technology is challenging and the huge success of Microsoft Kinect has been attracting researchers to uncover the potential of using this technology into computer vision applications. In this thesis, an upper-body motion analysis in a home-based system for stroke rehabilitation using novel RGB-D camera - Kinect is presented. We address this problem by first conducting a systematic analysis of the usability of Kinect for motion analysis in stroke rehabilitation. Then a hybrid upper body tracking approach is proposed which combines off-the-shelf skeleton tracking with a novel depth-fused mean shift tracking method. We proposed several kinematic features reliably extracted from the proposed inexpensive and portable motion capture system and classifiers that correlate torso movement to clinical measures of unimpaired and impaired. Experiment results show that the proposed sensing and analysis works reliably on measuring torso movement quality and is promising for end-point tracking. The system is currently being deployed for large-scale evaluations.
ContributorsDu, Tingfang (Author) / Turaga, Pavan (Thesis advisor) / Spanias, Andreas (Committee member) / Rikakis, Thanassis (Committee member) / Arizona State University (Publisher)
Created2012
Description
As a promising solution to the problem of acquiring and storing large amounts of image and video data, spatial-multiplexing camera architectures have received lot of attention in the recent past. Such architectures have the attractive feature of combining a two-step process of acquisition and compression of pixel measurements in a conventional camera, into a single step. A popular variant is the single-pixel camera that obtains measurements of the scene using a pseudo-random measurement matrix. Advances in compressive sensing (CS) theory in the past decade have supplied the tools that, in theory, allow near-perfect reconstruction of an image from these measurements even for sub-Nyquist sampling rates. However, current state-of-the-art reconstruction algorithms suffer from two drawbacks -- They are (1) computationally very expensive and (2) incapable of yielding high fidelity reconstructions for high compression ratios. In computer vision, the final goal is usually to perform an inference task using the images acquired and not signal recovery. With this motivation, this thesis considers the possibility of inference directly from compressed measurements, thereby obviating the need to use expensive reconstruction algorithms. It is often the case that non-linear features are used for inference tasks in computer vision. However, currently, it is unclear how to extract such features from compressed measurements. Instead, using the theoretical basis provided by the Johnson-Lindenstrauss lemma, discriminative features using smashed correlation filters are derived and it is shown that it is indeed possible to perform reconstruction-free inference at high compression ratios with only a marginal loss in accuracy. As a specific inference problem in computer vision, face recognition is considered, mainly beyond the visible spectrum such as in the short wave infra-red region (SWIR), where sensors are expensive.
ContributorsLohit, Suhas Anand (Author) / Turaga, Pavan (Thesis advisor) / Spanias, Andreas (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2015
Description
Computer Vision as a eld has gone through signicant changes in the last decade.
The eld has seen tremendous success in designing learning systems with hand-crafted
features and in using representation learning to extract better features. In this dissertation
some novel approaches to representation learning and task learning are studied.
Multiple-instance learning which is generalization of supervised learning, is one
example of task learning that is discussed. In particular, a novel non-parametric k-
NN-based multiple-instance learning is proposed, which is shown to outperform other
existing approaches. This solution is applied to a diabetic retinopathy pathology
detection problem eectively.
In cases of representation learning, generality of neural features are investigated
rst. This investigation leads to some critical understanding and results in feature
generality among datasets. The possibility of learning from a mentor network instead
of from labels is then investigated. Distillation of dark knowledge is used to eciently
mentor a small network from a pre-trained large mentor network. These studies help
in understanding representation learning with smaller and compressed networks.
The eld has seen tremendous success in designing learning systems with hand-crafted
features and in using representation learning to extract better features. In this dissertation
some novel approaches to representation learning and task learning are studied.
Multiple-instance learning which is generalization of supervised learning, is one
example of task learning that is discussed. In particular, a novel non-parametric k-
NN-based multiple-instance learning is proposed, which is shown to outperform other
existing approaches. This solution is applied to a diabetic retinopathy pathology
detection problem eectively.
In cases of representation learning, generality of neural features are investigated
rst. This investigation leads to some critical understanding and results in feature
generality among datasets. The possibility of learning from a mentor network instead
of from labels is then investigated. Distillation of dark knowledge is used to eciently
mentor a small network from a pre-trained large mentor network. These studies help
in understanding representation learning with smaller and compressed networks.
ContributorsVenkatesan, Ragav (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Yang, Yezhou (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2017
Description
The performance of most of the visual computing tasks depends on the quality of the features extracted from the raw data. Insightful feature representation increases the performance of many learning algorithms by exposing the underlying explanatory factors of the output for the unobserved input. A good representation should also handle anomalies in the data such as missing samples and noisy input caused by the undesired, external factors of variation. It should also reduce the data redundancy. Over the years, many feature extraction processes have been invented to produce good representations of raw images and videos.
The feature extraction processes can be categorized into three groups. The first group contains processes that are hand-crafted for a specific task. Hand-engineering features requires the knowledge of domain experts and manual labor. However, the feature extraction process is interpretable and explainable. Next group contains the latent-feature extraction processes. While the original feature lies in a high-dimensional space, the relevant factors for a task often lie on a lower dimensional manifold. The latent-feature extraction employs hidden variables to expose the underlying data properties that cannot be directly measured from the input. Latent features seek a specific structure such as sparsity or low-rank into the derived representation through sophisticated optimization techniques. The last category is that of deep features. These are obtained by passing raw input data with minimal pre-processing through a deep network. Its parameters are computed by iteratively minimizing a task-based loss.
In this dissertation, I present four pieces of work where I create and learn suitable data representations. The first task employs hand-crafted features to perform clinically-relevant retrieval of diabetic retinopathy images. The second task uses latent features to perform content-adaptive image enhancement. The third task ranks a pair of images based on their aestheticism. The goal of the last task is to capture localized image artifacts in small datasets with patch-level labels. For both these tasks, I propose novel deep architectures and show significant improvement over the previous state-of-art approaches. A suitable combination of feature representations augmented with an appropriate learning approach can increase performance for most visual computing tasks.
The feature extraction processes can be categorized into three groups. The first group contains processes that are hand-crafted for a specific task. Hand-engineering features requires the knowledge of domain experts and manual labor. However, the feature extraction process is interpretable and explainable. Next group contains the latent-feature extraction processes. While the original feature lies in a high-dimensional space, the relevant factors for a task often lie on a lower dimensional manifold. The latent-feature extraction employs hidden variables to expose the underlying data properties that cannot be directly measured from the input. Latent features seek a specific structure such as sparsity or low-rank into the derived representation through sophisticated optimization techniques. The last category is that of deep features. These are obtained by passing raw input data with minimal pre-processing through a deep network. Its parameters are computed by iteratively minimizing a task-based loss.
In this dissertation, I present four pieces of work where I create and learn suitable data representations. The first task employs hand-crafted features to perform clinically-relevant retrieval of diabetic retinopathy images. The second task uses latent features to perform content-adaptive image enhancement. The third task ranks a pair of images based on their aestheticism. The goal of the last task is to capture localized image artifacts in small datasets with patch-level labels. For both these tasks, I propose novel deep architectures and show significant improvement over the previous state-of-art approaches. A suitable combination of feature representations augmented with an appropriate learning approach can increase performance for most visual computing tasks.
ContributorsChandakkar, Parag Shridhar (Author) / Li, Baoxin (Thesis advisor) / Yang, Yezhou (Committee member) / Turaga, Pavan (Committee member) / Davulcu, Hasan (Committee member) / Arizona State University (Publisher)
Created2017
Description
Generating real-world content for VR is challenging in terms of capturing and processing at high resolution and high frame-rates. The content needs to represent a truly immersive experience, where the user can look around in 360-degree view and perceive the depth of the scene. The existing solutions only capture and offload the compute load to the server. But offloading large amounts of raw camera feeds takes longer latencies and poses difficulties for real-time applications. By capturing and computing on the edge, we can closely integrate the systems and optimize for low latency. However, moving the traditional stitching algorithms to battery constrained device needs at least three orders of magnitude reduction in power. We believe that close integration of capture and compute stages will lead to reduced overall system power.
We approach the problem by building a hardware prototype and characterize the end-to-end system bottlenecks of power and performance. The prototype has 6 IMX274 cameras and uses Nvidia Jetson TX2 development board for capture and computation. We found that capturing is bottlenecked by sensor power and data-rates across interfaces, whereas compute is limited by the total number of computations per frame. Our characterization shows that redundant capture and redundant computations lead to high power, huge memory footprint, and high latency. The existing systems lack hardware-software co-design aspects, leading to excessive data transfers across the interfaces and expensive computations within the individual subsystems. Finally, we propose mechanisms to optimize the system for low power and low latency. We emphasize the importance of co-design of different subsystems to reduce and reuse the data. For example, reusing the motion vectors of the ISP stage reduces the memory footprint of the stereo correspondence stage. Our estimates show that pipelining and parallelization on custom FPGA can achieve real time stitching.
We approach the problem by building a hardware prototype and characterize the end-to-end system bottlenecks of power and performance. The prototype has 6 IMX274 cameras and uses Nvidia Jetson TX2 development board for capture and computation. We found that capturing is bottlenecked by sensor power and data-rates across interfaces, whereas compute is limited by the total number of computations per frame. Our characterization shows that redundant capture and redundant computations lead to high power, huge memory footprint, and high latency. The existing systems lack hardware-software co-design aspects, leading to excessive data transfers across the interfaces and expensive computations within the individual subsystems. Finally, we propose mechanisms to optimize the system for low power and low latency. We emphasize the importance of co-design of different subsystems to reduce and reuse the data. For example, reusing the motion vectors of the ISP stage reduces the memory footprint of the stereo correspondence stage. Our estimates show that pipelining and parallelization on custom FPGA can achieve real time stitching.
ContributorsGunnam, Sridhar (Author) / LiKamWa, Robert (Thesis advisor) / Turaga, Pavan (Committee member) / Jayasuriya, Suren (Committee member) / Arizona State University (Publisher)
Created2018
Description
Mixture of experts is a machine learning ensemble approach that consists of individual models that are trained to be ``experts'' on subsets of the data, and a gating network that provides weights to output a combination of the expert predictions. Mixture of experts models do not currently see wide use due to difficulty in training diverse experts and high computational requirements. This work presents modifications of the mixture of experts formulation that use domain knowledge to improve training, and incorporate parameter sharing among experts to reduce computational requirements.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
First, this work presents an application of mixture of experts models for quality robust visual recognition. First it is shown that human subjects outperform deep neural networks on classification of distorted images, and then propose a model, MixQualNet, that is more robust to distortions. The proposed model consists of ``experts'' that are trained on a particular type of image distortion. The final output of the model is a weighted sum of the expert models, where the weights are determined by a separate gating network. The proposed model also incorporates weight sharing to reduce the number of parameters, as well as increase performance.
Second, an application of mixture of experts to predict visual saliency is presented. A computational saliency model attempts to predict where humans will look in an image. In the proposed model, each expert network is trained to predict saliency for a set of closely related images. The final saliency map is computed as a weighted mixture of the expert networks' outputs, with weights determined by a separate gating network. The proposed model achieves better performance than several other visual saliency models and a baseline non-mixture model.
Finally, this work introduces a saliency model that is a weighted mixture of models trained for different levels of saliency. Levels of saliency include high saliency, which corresponds to regions where almost all subjects look, and low saliency, which corresponds to regions where some, but not all subjects look. The weighted mixture shows improved performance compared with baseline models because of the diversity of the individual model predictions.
ContributorsDodge, Samuel Fuller (Author) / Karam, Lina (Thesis advisor) / Jayasuriya, Suren (Committee member) / Li, Baoxin (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2018
Description
Non-line-of-sight (NLOS) imaging of objects not visible to either the camera or illumina-
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
tion source is a challenging task with vital applications including surveillance and robotics.
Recent NLOS reconstruction advances have been achieved using time-resolved measure-
ments. Acquiring these time-resolved measurements requires expensive and specialized
detectors and laser sources. In work proposes a data-driven approach for NLOS 3D local-
ization requiring only a conventional camera and projector. The localisation is performed
using a voxelisation and a regression problem. Accuracy of greater than 90% is achieved
in localizing a NLOS object to a 5cm × 5cm × 5cm volume in real data. By adopting
the regression approach an object of width 10cm to localised to approximately 1.5cm. To
generalize to line-of-sight (LOS) scenes with non-planar surfaces, an adaptive lighting al-
gorithm is adopted. This algorithm, based on radiosity, identifies and illuminates scene
patches in the LOS which most contribute to the NLOS light paths, and can factor in sys-
tem power constraints. Improvements ranging from 6%-15% in accuracy with a non-planar
LOS wall using adaptive lighting is reported, demonstrating the advantage of combining
the physics of light transport with active illumination for data-driven NLOS imaging.
ContributorsChandran, Sreenithy (Author) / Jayasuriya, Suren (Thesis advisor) / Turaga, Pavan (Committee member) / Dasarathy, Gautam (Committee member) / Arizona State University (Publisher)
Created2019
Description
Multi-sensor fusion is a fundamental problem in Robot Perception. For a robot to operate in a real world environment, multiple sensors are often needed. Thus, fusing data from various sensors accurately is vital for robot perception. In the first part of this thesis, the problem of fusing information from a LIDAR, a color camera and a thermal camera to build RGB-Depth-Thermal (RGBDT) maps is investigated. An algorithm that solves a non-linear optimization problem to compute the relative pose between the cameras and the LIDAR is presented. The relative pose estimate is then used to find the color and thermal texture of each LIDAR point. Next, the various sources of error that can cause the mis-coloring of a LIDAR point after the cross- calibration are identified. Theoretical analyses of these errors reveal that the coloring errors due to noisy LIDAR points, errors in the estimation of the camera matrix, and errors in the estimation of translation between the sensors disappear with distance. But errors in the estimation of the rotation between the sensors causes the coloring error to increase with distance.
On a robot (vehicle) with multiple sensors, sensor fusion algorithms allow us to represent the data in the vehicle frame. But data acquired temporally in the vehicle frame needs to be registered in a global frame to obtain a map of the environment. Mapping techniques involving the Iterative Closest Point (ICP) algorithm and the Normal Distributions Transform (NDT) assume that a good initial estimate of the transformation between the 3D scans is available. This restricts the ability to stitch maps that were acquired at different times. Mapping can become flexible if maps that were acquired temporally can be merged later. To this end, the second part of this thesis focuses on developing an automated algorithm that fuses two maps by finding a congruent set of five points forming a pyramid.
Mapping has various application domains beyond Robot Navigation. The third part of this thesis considers a unique application domain where the surface displace- ments caused by an earthquake are to be recovered using pre- and post-earthquake LIDAR data. A technique to recover the 3D surface displacements is developed and the results are presented on real earthquake datasets: El Mayur Cucupa earthquake, Mexico, 2010 and Fukushima earthquake, Japan, 2011.
On a robot (vehicle) with multiple sensors, sensor fusion algorithms allow us to represent the data in the vehicle frame. But data acquired temporally in the vehicle frame needs to be registered in a global frame to obtain a map of the environment. Mapping techniques involving the Iterative Closest Point (ICP) algorithm and the Normal Distributions Transform (NDT) assume that a good initial estimate of the transformation between the 3D scans is available. This restricts the ability to stitch maps that were acquired at different times. Mapping can become flexible if maps that were acquired temporally can be merged later. To this end, the second part of this thesis focuses on developing an automated algorithm that fuses two maps by finding a congruent set of five points forming a pyramid.
Mapping has various application domains beyond Robot Navigation. The third part of this thesis considers a unique application domain where the surface displace- ments caused by an earthquake are to be recovered using pre- and post-earthquake LIDAR data. A technique to recover the 3D surface displacements is developed and the results are presented on real earthquake datasets: El Mayur Cucupa earthquake, Mexico, 2010 and Fukushima earthquake, Japan, 2011.
ContributorsKrishnan, Aravindhan K (Author) / Saripalli, Srikanth (Thesis advisor) / Klesh, Andrew (Committee member) / Fainekos, Georgios (Committee member) / Thangavelautham, Jekan (Committee member) / Turaga, Pavan (Committee member) / Arizona State University (Publisher)
Created2016
Description
High-level inference tasks in video applications such as recognition, video retrieval, and zero-shot classification have become an active research area in recent years. One fundamental requirement for such applications is to extract high-quality features that maintain high-level information in the videos.
Many video feature extraction algorithms have been purposed, such as STIP, HOG3D, and Dense Trajectories. These algorithms are often referred to as “handcrafted” features as they were deliberately designed based on some reasonable considerations. However, these algorithms may fail when dealing with high-level tasks or complex scene videos. Due to the success of using deep convolution neural networks (CNNs) to extract global representations for static images, researchers have been using similar techniques to tackle video contents. Typical techniques first extract spatial features by processing raw images using deep convolution architectures designed for static image classifications. Then simple average, concatenation or classifier-based fusion/pooling methods are applied to the extracted features. I argue that features extracted in such ways do not acquire enough representative information since videos, unlike images, should be characterized as a temporal sequence of semantically coherent visual contents and thus need to be represented in a manner considering both semantic and spatio-temporal information.
In this thesis, I propose a novel architecture to learn semantic spatio-temporal embedding for videos to support high-level video analysis. The proposed method encodes video spatial and temporal information separately by employing a deep architecture consisting of two channels of convolutional neural networks (capturing appearance and local motion) followed by their corresponding Fully Connected Gated Recurrent Unit (FC-GRU) encoders for capturing longer-term temporal structure of the CNN features. The resultant spatio-temporal representation (a vector) is used to learn a mapping via a Fully Connected Multilayer Perceptron (FC-MLP) to the word2vec semantic embedding space, leading to a semantic interpretation of the video vector that supports high-level analysis. I evaluate the usefulness and effectiveness of this new video representation by conducting experiments on action recognition, zero-shot video classification, and semantic video retrieval (word-to-video) retrieval, using the UCF101 action recognition dataset.
Many video feature extraction algorithms have been purposed, such as STIP, HOG3D, and Dense Trajectories. These algorithms are often referred to as “handcrafted” features as they were deliberately designed based on some reasonable considerations. However, these algorithms may fail when dealing with high-level tasks or complex scene videos. Due to the success of using deep convolution neural networks (CNNs) to extract global representations for static images, researchers have been using similar techniques to tackle video contents. Typical techniques first extract spatial features by processing raw images using deep convolution architectures designed for static image classifications. Then simple average, concatenation or classifier-based fusion/pooling methods are applied to the extracted features. I argue that features extracted in such ways do not acquire enough representative information since videos, unlike images, should be characterized as a temporal sequence of semantically coherent visual contents and thus need to be represented in a manner considering both semantic and spatio-temporal information.
In this thesis, I propose a novel architecture to learn semantic spatio-temporal embedding for videos to support high-level video analysis. The proposed method encodes video spatial and temporal information separately by employing a deep architecture consisting of two channels of convolutional neural networks (capturing appearance and local motion) followed by their corresponding Fully Connected Gated Recurrent Unit (FC-GRU) encoders for capturing longer-term temporal structure of the CNN features. The resultant spatio-temporal representation (a vector) is used to learn a mapping via a Fully Connected Multilayer Perceptron (FC-MLP) to the word2vec semantic embedding space, leading to a semantic interpretation of the video vector that supports high-level analysis. I evaluate the usefulness and effectiveness of this new video representation by conducting experiments on action recognition, zero-shot video classification, and semantic video retrieval (word-to-video) retrieval, using the UCF101 action recognition dataset.
ContributorsHu, Sheng-Hung (Author) / Li, Baoxin (Thesis advisor) / Turaga, Pavan (Committee member) / Liang, Jianming (Committee member) / Tong, Hanghang (Committee member) / Arizona State University (Publisher)
Created2016
Description
Compressive sensing theory allows to sense and reconstruct signals/images with lower sampling rate than Nyquist rate. Applications in resource constrained environment stand to benefit from this theory, opening up many possibilities for new applications at the same time. The traditional inference pipeline for computer vision sequence reconstructing the image from compressive measurements. However,the reconstruction process is a computationally expensive step that also provides poor results at high compression rate. There have been several successful attempts to perform inference tasks directly on compressive measurements such as activity recognition. In this thesis, I am interested to tackle a more challenging vision problem - Visual question answering (VQA) without reconstructing the compressive images. I investigate the feasibility of this problem with a series of experiments, and I evaluate proposed methods on a VQA dataset and discuss promising results and direction for future work.
ContributorsHuang, Li-Chin (Author) / Turaga, Pavan (Thesis advisor) / Yang, Yezhou (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2017