Matching Items (1,629)
Filtering by

Clear all filters

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

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.
ContributorsGunnam, Sridhar (Author) / LiKamWa, Robert (Thesis advisor) / Turaga, Pavan (Committee member) / Jayasuriya, Suren (Committee member) / Arizona State University (Publisher)
Created2018
156747-Thumbnail Image.png
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

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.
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
157215-Thumbnail Image.png
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

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.
ContributorsChandran, Sreenithy (Author) / Jayasuriya, Suren (Thesis advisor) / Turaga, Pavan (Committee member) / Dasarathy, Gautam (Committee member) / Arizona State University (Publisher)
Created2019
157065-Thumbnail Image.png
Description
The detection and segmentation of objects appearing in a natural scene, often referred to as Object Detection, has gained a lot of interest in the computer vision field. Although most existing object detectors aim to detect all the objects in a given scene, it is important to evaluate whether these

The detection and segmentation of objects appearing in a natural scene, often referred to as Object Detection, has gained a lot of interest in the computer vision field. Although most existing object detectors aim to detect all the objects in a given scene, it is important to evaluate whether these methods are capable of detecting the salient objects in the scene when constraining the number of proposals that can be generated due to constraints on timing or computations during execution. Salient objects are objects that tend to be more fixated by human subjects. The detection of salient objects is important in applications such as image collection browsing, image display on small devices, and perceptual compression.

This thesis proposes a novel evaluation framework that analyses the performance of popular existing object proposal generators in detecting the most salient objects. This work also shows that, by incorporating saliency constraints, the number of generated object proposals and thus the computational cost can be decreased significantly for a target true positive detection rate (TPR).

As part of the proposed framework, salient ground-truth masks are generated from the given original ground-truth masks for a given dataset. Given an object detection dataset, this work constructs salient object location ground-truth data, referred to here as salient ground-truth data for short, that only denotes the locations of salient objects. This is obtained by first computing a saliency map for the input image and then using it to assign a saliency score to each object in the image. Objects whose saliency scores are sufficiently high are referred to as salient objects. The detection rates are analyzed for existing object proposal generators with respect to the original ground-truth masks and the generated salient ground-truth masks.

As part of this work, a salient object detection database with salient ground-truth masks was constructed from the PASCAL VOC 2007 dataset. Not only does this dataset aid in analyzing the performance of existing object detectors for salient object detection, but it also helps in the development of new object detection methods and evaluating their performance in terms of successful detection of salient objects.
ContributorsKotamraju, Sai Prajwal (Author) / Karam, Lina J (Thesis advisor) / Yu, Hongbin (Committee member) / Jayasuriya, Suren (Committee member) / Arizona State University (Publisher)
Created2019
133352-Thumbnail Image.png
Description
The inherent risk in testing drugs has been hotly debated since the government first started regulating the drug industry in the early 1900s. Who can assume the risks associated with trying new pharmaceuticals is unclear when looked at through society's lens. In the mid twentieth century, the US Food and

The inherent risk in testing drugs has been hotly debated since the government first started regulating the drug industry in the early 1900s. Who can assume the risks associated with trying new pharmaceuticals is unclear when looked at through society's lens. In the mid twentieth century, the US Food and Drug Administration (FDA) published several guidance documents encouraging researchers to exclude women from early clinical drug research. The motivation to publish those documents and the subsequent guidance documents in which the FDA and other regulatory offices established their standpoints on women in drug research may have been connected to current events at the time. The problem of whether women should be involved in drug research is a question of who can assume risk and who is responsible for disseminating what specific kinds of information. The problem tends to be framed as one that juxtaposes the health of women and fetuses and sets their health as in opposition. That opposition, coupled with the inherent uncertainty in testing drugs, provides for a complex set of issues surrounding consent and access to information.
ContributorsMeek, Caroline Jane (Author) / Maienschein, Jane (Thesis director) / Brian, Jennifer (Committee member) / School of Life Sciences (Contributor) / Sanford School of Social and Family Dynamics (Contributor) / Barrett, The Honors College (Contributor)
Created2018-05
131502-Thumbnail Image.png
Description
Social-emotional learning (SEL) methods are beginning to receive global attention in primary school education, yet the dominant emphasis on implementing these curricula is in high-income, urbanized areas. Consequently, the unique features of developing and integrating such methods in middle- or low-income rural areas are unclear. Past studies suggest that students

Social-emotional learning (SEL) methods are beginning to receive global attention in primary school education, yet the dominant emphasis on implementing these curricula is in high-income, urbanized areas. Consequently, the unique features of developing and integrating such methods in middle- or low-income rural areas are unclear. Past studies suggest that students exposed to SEL programs show an increase in academic performance, improved ability to cope with stress, and better attitudes about themselves, others, and school, but these curricula are designed with an urban focus. The purpose of this study was to conduct a needs-based analysis to investigate components specific to a SEL curriculum contextualized to rural primary schools. A promising organization committed to rural educational development is Barefoot College, located in Tilonia, Rajasthan, India. In partnership with Barefoot, we designed an ethnographic study to identify and describe what teachers and school leaders consider the highest needs related to their students' social and emotional education. To do so, we interviewed 14 teachers and school leaders individually or in a focus group to explore their present understanding of “social-emotional learning” and the perception of their students’ social and emotional intelligence. Analysis of this data uncovered common themes among classroom behaviors and prevalent opportunities to address social and emotional well-being among students. These themes translated into the three overarching topics and eight sub-topics explored throughout the curriculum, and these opportunities guided the creation of the 21 modules within it. Through a design-based research methodology, we developed a 40-hour curriculum by implementing its various modules within seven Barefoot classrooms alongside continuous reiteration based on teacher feedback and participant observation. Through this process, we found that student engagement increased during contextualized SEL lessons as opposed to traditional methods. In addition, we found that teachers and students preferred and performed better with an activities-based approach. These findings suggest that rural educators must employ particular teaching strategies when addressing SEL, including localized content and an experiential-learning approach. Teachers reported that as their approach to SEL shifted, they began to unlock the potential to build self-aware, globally-minded students. This study concludes that social and emotional education cannot be treated in a generalized manner, as curriculum development is central to the teaching-learning process.
ContributorsBucker, Delaney Sue (Author) / Carrese, Susan (Thesis director) / Barab, Sasha (Committee member) / School of Life Sciences (Contributor, Contributor) / School of Civic & Economic Thought and Leadership (Contributor) / School of International Letters and Cultures (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
131507-Thumbnail Image.png
Description
As of 2019, 30 US states have adopted abortion-specific informed consent laws that require state health departments to develop and disseminate written informational materials to patients seeking an abortion. Abortion is the only medical procedure for which states dictate the content of informed consent counseling. State abortion counseling materials have

As of 2019, 30 US states have adopted abortion-specific informed consent laws that require state health departments to develop and disseminate written informational materials to patients seeking an abortion. Abortion is the only medical procedure for which states dictate the content of informed consent counseling. State abortion counseling materials have been criticized for containing inaccurate and misleading information, but overall, informed consent laws for abortion do not often receive national attention. The objective of this project was to determine the importance of informed consent laws to achieving the larger goal of dismantling the right to abortion. I found that informed consent counseling materials in most states contain a full timeline of fetal development, along with information about the risks of abortion, the risks of childbirth, and alternatives to abortion. In addition, informed consent laws for abortion are based on model legislation called the “Women’s Right to Know Act” developed by Americans United for Life (AUL). AUL calls itself the legal architect of the pro-life movement and works to pass laws at the state level that incrementally restrict abortion access so that it gradually becomes more difficult to exercise the right to abortion established by Roe v. Wade. The “Women’s Right to Know Act” is part of a larger package of model legislation called the “Women’s Protection Project,” a cluster of laws that place restrictions on abortion providers, purportedly to protect women, but actually to decrease abortion access. “Women’s Right to Know” counseling laws do not directly deny access to abortion, but they do reinforce key ideas important to the anti-abortion movement, like the concept of fetal personhood, distrust in medical professionals, the belief that pregnant people cannot be fully autonomous individuals, and the belief that abortion is not an ordinary medical procedure and requires special government oversight. “Women’s Right to Know” laws use the language of informed consent and the purported goal of protecting women to legitimize those ideas, and in doing so, they significantly undermine the right to abortion. The threat to abortion rights posed by laws like the “Women’s Right to Know” laws indicates the need to reevaluate and strengthen our ethical defense of the right to abortion.
ContributorsVenkatraman, Richa (Author) / Maienschein, Jane (Thesis director) / Brian, Jennifer (Thesis director) / Abboud, Carolina (Committee member) / Historical, Philosophical & Religious Studies (Contributor) / School of Life Sciences (Contributor, Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
131521-Thumbnail Image.png
Description
Turbidity is a known problem for UV water treatment systems as suspended particles can shield contaminants from the UV radiation. UV systems that utilize a reflective radiation chamber may be able to decrease the impact of turbidity on the efficacy of the system. The purpose of this study was to

Turbidity is a known problem for UV water treatment systems as suspended particles can shield contaminants from the UV radiation. UV systems that utilize a reflective radiation chamber may be able to decrease the impact of turbidity on the efficacy of the system. The purpose of this study was to determine how kaolin clay and gram flour turbidity affects inactivation of Escherichia coli (E. coli) when using a UV system with a reflective chamber. Both sources of turbidity were shown to reduce the inactivation of E. coli with increasing concentrations. Overall, it was shown that increasing kaolin clay turbidity had a consistent effect on reducing UV inactivation across UV doses. Log inactivation was reduced by 1.48 log for the low UV dose and it was reduced by at least 1.31 log for the low UV dose. Gram flour had a similar effect to the clay at the lower UV dose, reducing log inactivation by 1.58 log. At the high UV dose, there was no change in UV inactivation with an increase in turbidity. In conclusion, turbidity has a significant impact on the efficacy of UV disinfection. Therefore, removing turbidity from water is an essential process to enhance UV efficiency for the disinfection of microbial pathogens.
ContributorsMalladi, Rohith (Author) / Abbaszadegan, Morteza (Thesis director) / Alum, Absar (Committee member) / Fox, Peter (Committee member) / School of Human Evolution & Social Change (Contributor) / School of Life Sciences (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
131526-Thumbnail Image.png
Description
Aquatic macroinvertebrates are important for many ecological processes within river ecosystems and, as a result, their abundance and diversity are considered indicators of water quality and ecosystem health. Macroinvertebrates can be classified into functional feeding groups (FFG) based on morphological-behavioral adaptations. FFG ratios can shift due to changes

Aquatic macroinvertebrates are important for many ecological processes within river ecosystems and, as a result, their abundance and diversity are considered indicators of water quality and ecosystem health. Macroinvertebrates can be classified into functional feeding groups (FFG) based on morphological-behavioral adaptations. FFG ratios can shift due to changes in normal disturbance patterns, such as changes in precipitation, and from human impact. Due to their increased sensitivity to environmental changes, it has become more important to protect and monitor aquatic and riparian communities in arid regions as climate change continues to intensify. Therefore, the diversity and richness of macroinvertebrate FFGs before and after monsoon and winter storm seasons were analyzed to determine the effect of flow-related disturbances. Ecosystem size was also considered, as watershed area has been shown to affect macroinvertebrate diversity. There was no strong support for flow-related disturbance or ecosystem size on macroinvertebrate diversity and richness. This may indicate a need to explore other parameters of macroinvertebrate community assembly. Establishing how disturbance affects aquatic macroinvertebrate communities will provide a key understanding as to what the stream communities will look like in the future, as anthropogenic impacts continue to affect more vulnerable ecosystems.
ContributorsSainz, Ruby (Author) / Sabo, John (Thesis director) / Grimm, Nancy (Committee member) / Lupoli, Christina (Committee member) / School of Life Sciences (Contributor, Contributor) / Barrett, The Honors College (Contributor)
Created2020-05
131531-Thumbnail Image.png
Description
This study evaluates medical pluralism among 1.5 generation Indian American immigrants. 1.5 generation Indian Americans (N=16) were surveyed regarding their engagement in complementary and alternative medical systems (CAM), how immigration affected that, and reasons for and for not continuing the use of CAM. Results indicated most 1.5 Indian immigrants currently

This study evaluates medical pluralism among 1.5 generation Indian American immigrants. 1.5 generation Indian Americans (N=16) were surveyed regarding their engagement in complementary and alternative medical systems (CAM), how immigration affected that, and reasons for and for not continuing the use of CAM. Results indicated most 1.5 Indian immigrants currently engage in CAM, given that their parents also engage in CAM. The top reasons respondents indicated continued engagement in CAM was that it has no side effects and is preventative. Reasons for not practicing CAM included feeling out of place, not living with parents or not believing in CAM. After immigration, most participants decreased or stopped their engagement in CAM. More women than men continued to practice CAM after immigration. From the results, it was concluded that CAM is still important to 1.5 generation Indian immigrants.
ContributorsMurugesh, Subhiksha (Author) / Stotts, Rhian (Thesis director) / Mubayi, Anuj (Committee member) / School of Human Evolution & Social Change (Contributor) / School of Life Sciences (Contributor) / Barrett, The Honors College (Contributor)
Created2020-05