Engineering Commons^™

Evaluation Of Robust Deep Learning Pipelines Targeting Low Swap Edge Deployment, David Carter Cornett

Masters Theses

The deep learning technique of convolutional neural networks (CNNs) has greatly advanced the state-of-the-art for computer vision tasks such as image classification and object detection. These solutions rely on large systems leveraging wattage-hungry GPUs to provide the computational power to achieve such performance. However, the size, weight and power (SWaP) requirements of these conventional GPU-based deep learning systems are not suitable when a solution requires deployment to so called "Edge" environments such as autonomous vehicles, unmanned aerial vehicles (UAVs) and smart security cameras.

The objective of this work is to benchmark FPGA-based alternatives to conventional GPU systems that have the …

Analysis Of Hardware Accelerated Deep Learning And The Effects Of Degradation On Performance, Samuel C. Leach

Masters Theses

As convolutional neural networks become more prevalent in research and real world applications, the need for them to be faster and more robust will be a constant battle. This thesis investigates the effect of degradation being introduced to an image prior to object recognition with a convolutional neural network. As well as experimenting with methods to reduce the degradation and improve performance. Gaussian smoothing and additive Gaussian noise are both analyzed degradation models within this thesis and are reduced with Gaussian and Butterworth masks using unsharp masking and smoothing, respectively. The results show that each degradation is disruptive to the …

Approaches To Studying Bacterial Biofilms In The Bioeconomy With Nanofabrication Techniques And Engineered Platforms., Michelle Caroline Halsted

Doctoral Dissertations

Studies that estimate more than 90% of bacteria subsist in a biofilm state to survive environmental stressors. These biofilms persist on man-made and natural surfaces, and examples of the rich biofilm diversity extends from the roots of bioenergy crops to electroactive biofilms in bioelectrochemical reactors. Efforts to optimize microbial systems in the bioeconomy will benefit from an improved fundamental understanding of bacterial biofilms. An understanding of these microbial systems shows promise to increase crop yields with precision agriculture (e.g. biosynthetic fertilizer, microbial pesticides, and soil remediation) and increase commodity production yields in bioreactors. Yet conventional laboratory methods investigate these micron-scale …

Dynamic In Vivo Skeletal Feature Tracking Via Fluoroscopy Using A Human Gait Model, William Patrick Anderson

Doctoral Dissertations

The Tracking Fluoroscope System II, a mobile robotic fluoroscopy platform, developed and built at the University of Tennessee, Knoxville, presently employs a pattern matching algorithm in order to identify and track a marker placed upon a subject’s knee joint of interest. The purpose of this research is to generate a new tracking algorithm based around the human gait cycle for prediction and improving the overall accuracy of joint tracking.

This research centers around processing the acquired x-ray images of the desired knee joint obtained during standard clinical operation in order to identify and track directly through the acquired image. Due …

Architecture For Real-Time, Low-Swap Embedded Vision Using Fpgas, Steven Andrew Clukey

Masters Theses

In this thesis we designed, prototyped, and constructed a printed circuit board for real-time, low size, weight, and power (SWaP) HDMI video processing and developed a general purpose library of image processing functions for FPGAs.

The printed circuit board is a baseboard for a Xilinx Zynq based system-on-module (SoM). The board provides power, HDMI input, and HDMI output to the SoM and enables low-SWaP, high-resolution, real-time video processing.

The image processing library for FPGAs is designed for high performance and high reusability. These objectives are achieved by utilizing the Chisel hardware construction language to create parameterized modules that construct low-level …

Computational Analysis Of Neutron Scattering Data, Benjamin Walter Martin

Doctoral Dissertations

This work explores potential methods for use in the detection and classification of defects within crystal structures via analysis of diffuse scattering data generated by single crystal neutron scattering experiments. The proposed defect detection methodology uses machine learning and image processing techniques to perform image texture analysis on neutron diffraction patterns generated by neutron scattering simulations. Once the methodology is presented, it is tested via a series of defect detection problems of increasing difficulty which utilize neutron scattering data simulated by a number of simulation techniques. As the problem difficulty is increased, the defect detection methodology is refined in order …

An Expert System For Guitar Sheet Music To Guitar Tablature, Chuanjun He

Doctoral Dissertations

This project applies analysis, design and implementation of the Optical Music Recognition (OMR) to an expert system for transforming guitar sheet music to guitar tablature. The first part includes image processing and music semantic interpretation to interpret and transform sheet music or printed scores into editable and playable electronic form. Then after importing the electronic form of music into internal data structures, our application uses effective pruning to explore the entire search space to find the best guitar tablature. Also considered are alternate guitar tunings and transposition of the music to improve the resulting tablature.

Deep Machine Learning With Spatio-Temporal Inference, Thomas Paul Karnowski

Doctoral Dissertations

Deep Machine Learning (DML) refers to methods which utilize hierarchies of more than one or two layers of computational elements to achieve learning. DML may draw upon biomemetic models, or may be simply biologically-inspired. Regardless, these architectures seek to employ hierarchical processing as means of mimicking the ability of the human brain to process a myriad of sensory data and make meaningful decisions based on this data. In this dissertation we present a novel DML architecture which is biologically-inspired in that (1) all processing is performed hierarchically; (2) all processing units are identical; and (3) processing captures both spatial and …