Engineering Commons^™

Localization Using Convolutional Neural Networks, Shannon D. Fong

Computer Engineering

With the increased accessibility to powerful GPUs, ability to develop machine learning algorithms has increased significantly. Coupled with open source deep learning frameworks, average users are now able to experiment with convolutional neural networks (CNNs) to solve novel problems. This project sought to train a CNN capable of classifying between various locations within a building. A single continuous video was taken while standing at each desired location so that every class in the neural network was represented by a single video. Each location was given a number to be used for classification and the video was subsequently titled locX. These …

Automatic Identification Of Animals In The Wild: A Comparative Study Between C-Capsule Networks And Deep Convolutional Neural Networks., Joel Kamdem Teto, Ying Xie

Master of Science in Computer Science Theses

The evolution of machine learning and computer vision in technology has driven a lot of

improvements and innovation into several domains. We see it being applied for credit decisions, insurance quotes, malware detection, fraud detection, email composition, and any other area having enough information to allow the machine to learn patterns. Over the years the number of sensors, cameras, and cognitive pieces of equipment placed in the wilderness has been growing exponentially. However, the resources (human) to leverage these data into something meaningful are not improving at the same rate. For instance, a team of scientist volunteers took 8.4 years, …

Investigating Dataset Distinctiveness, Andrew Ulmer, Kent W. Gauen, Yung-Hsiang Lu, Zohar R. Kapach, Daniel P. Merrick

The Summer Undergraduate Research Fellowship (SURF) Symposium

Just as a human might struggle to interpret another human’s handwriting, a computer vision program might fail when asked to perform one task in two different domains. To be more specific, visualize a self-driving car as a human driver who had only ever driven on clear, sunny days, during daylight hours. This driver – the self-driving car – would inevitably face a significant challenge when asked to drive when it is violently raining or foggy during the night, putting the safety of its passengers in danger. An extensive understanding of the data we use to teach computer vision models – …

Hierarchical Bayesian Data Fusion Using Autoencoders, Yevgeniy Vladimirovich Reznichenko

Master's Theses (2009 -)

In this thesis, a novel method for tracker fusion is proposed and evaluated for vision-based tracking. This work combines three distinct popular techniques into a recursive Bayesian estimation algorithm. First, semi supervised learning approaches are used to partition data and to train a deep neural network that is capable of capturing normal visual tracking operation and is able to detect anomalous data. We compare various methods by examining their respective receiver operating conditions (ROC) curves, which represent the trade off between specificity and sensitivity for various detection threshold levels. Next, we incorporate the trained neural networks into an existing data …

Lionfish Detection System, Carmelo Furlan, Andrew Boniface

Computer Engineering

Deep neural networks have proven to be an effective method in classification of images. The ability to recognize objects has opened the door for many new systems which use image classification to solve challenging problems where conventional image classification would be inadequate. We trained a large, deep convolutional neural network to identify lionfish from other species that might be found in the same habitats. Google’s Inception framework served as a powerful platform for our fish recognition system. By using transfer learning, we were able to obtain exceptional results for the classification of different species of fish. The convolutional neural network …

Corridor Navigation For Monocular Vision Mobile Robots, Matthew James Ng

Master's Theses

Monocular vision robots use a single camera to process information about its environment. By analyzing this scene, the robot can determine the best navigation direction. Many modern approaches to robot hallway navigation involve using a plethora of sensors to detect certain features in the environment. This can be laser range finders, inertial measurement units, motor encoders, and cameras.

By combining all these sensors, there is unused data which could be useful for navigation. To draw back and develop a baseline approach, this thesis explores the reliability and capability of solely using a camera for navigation. The basic navigation structure begins …

Automated Pruning Of Greenhouse Indeterminate Tomato Plants, Joey M. Angeja

Master's Theses

Pruning of indeterminate tomato plants is vital for a profitable yield and it still remains a manual process. There has been research in automated pruning of grapevines, trees, and other plants, but tomato plants have yet to be explored. Wage increases are contributing to the depleting profits of greenhouse tomato farmers. Rises in population are the driving force behind the need for efficient growing techniques. The major contribution of this thesis is a computer vision algorithm for detecting greenhouse tomato pruning points without the use of depth sensors. Given an up-close 2-D image of a tomato stem with the background …

Mitigating The Effects Of Boom Occlusion On Automated Aerial Refueling Through Shadow Volumes, Zachary C. Paulson

Theses and Dissertations

In flight refueling of Unmanned Aerial Vehicles (UAVs) is critical to the United States Air Force (USAF). However, the large communication latency between a ground-based operator and his/her remote UAV makes docking with a refueling tanker unsafe. This latency may be mitigated by leveraging a tanker-centric stereo vision system. The vision system observes and computes an approaching receiver's relative position and orientation offering a low-latency, high frequency docking solution. Unfortunately, the boom -- an articulated refueling arm responsible for physically pumping fuel into the receiver -- occludes large portions of the receiver especially as the receiver approaches and docks with …

Stereo Vision: A Comparison Of Synthetic Imagery Vs. Real World Imagery For The Automated Aerial Refueling Problem, Nicholas J. Seydel

Theses and Dissertations

Missions using unmanned aerial vehicles have increased in the past decade. Currently, there is no way to refuel these aircraft. Accomplishing automated aerial refueling can be made possible using the stereo vision system on a tanker. Real world experiments for the automated aerial refueling problem are expensive and time consuming. Currently, simulations performed in a virtual world have shown promising results using computer vision. It is possible to use the virtual world as a substitute environment for the real world. This research compares the performance of stereo vision algorithms on synthetic and real world imagery.

Collaborative Appearance-Based Place Recognition And Improving Place Recognition Using Detection Of Dynamic Objects, Juan Pablo Munoz

Dissertations, Theses, and Capstone Projects

This dissertation makes contributions to the problem of Long-Term Appearance-Based Place Recognition. We present a framework for place recognition in a collaborative scheme and a method to reduce the impact of dynamic objects on place representations. We demonstrate our findings using a state-of-the-art place recognition approach.

We begin in Part I by describing the general problem of place recognition and its importance in applications where accurate localization is crucial. We discuss feature detection and description and also explain the functioning of several place recognition frameworks.

In Part II, we present a novel framework for collaboration between agents from a pure …

Fully Transparent Computer Vision Framework For Ship Detection And Tracking In Satellite Imagery, Jason T. Gottweis

Browse all Theses and Dissertations

Tracking of ships in satellite imagery is a challenging problem in remote sensing since it requires both object detection and object recognition. Most of the resources available only cover one of these problems and are often filled with machine learning techniques which are costly to train. Additionally, the techniques covered in these resources are often difficult to replicate or may be hard to combine with other solutions to get a full tracking algorithm. The proposed framework offers a transparent and efficient alternative to machine learning approaches and includes preprocessing, detection, and recognition needed for tracking. All components of the framework …

Investigating The Application Of Deep Convolutional Neural Networks In Semi-Supervised Video Object Segmentation, Jayadeep Sasikumar

Dissertations

This thesis investigates the different approaches to video object segmentation and the current state-of-the-art in the discipline, focusing on the different deep learning techniques used to solve the problem. The primary contribution of the thesis is the investigation of usefulness of Exponential Linear Units as activation functions for deep convolutional neural architectures trained to perform object semi-supervised segmentation in videos. Mask R-CNN was chosen as the base convolutional neural architecture, with the view of extending the image segmentation algorithm to videos. Two models were created, one with Rectified Linear Units and the other with Exponential Linear Units as the respective …