Physical Sciences and Mathematics Commons^™

A Real-Time 3d Object Detection, Recognition And Presentation System On A Mobile Device For Assistive Navigation, Jin Chen

Dissertations and Theses

This thesis proposes an integrated solution for 3D object detection, recognition, and presentation to increase accessibility for various user groups in indoor areas through a mobile application. The system has three major components: a 3D object detection module, an object tracking and update module, and a voice and AR-enhanced interface. The 3D object detection module consists of pre-trained 2D object detectors and 3D bounding box estimation methods to detect the 3D poses and sizes of the objects in each camera frame. This module can easily adapt to various 2D object detectors (e.g., YOLO, SSD, Mask RCNN) based on the requested …

A Citizen-Science Approach For Urban Flood Risk Analysis Using Data Science And Machine Learning, Candace Agonafir

Dissertations and Theses

Street flooding is problematic in urban areas, where impervious surfaces, such as concrete, brick, and asphalt prevail, impeding the infiltration of water into the ground. During rain events, water ponds and rise to levels that cause considerable economic damage and physical harm. The main goal of this dissertation is to develop novel approaches toward the comprehension of urban flood risk using data science techniques on crowd-sourced data. This is accomplished by developing a series of data-driven models to identify flood factors of significance and localized areas of flood vulnerability in New York City (NYC). First, the infrastructural (catch basin clogs, …

Multimodal Data Integration For Real-Time Indoor Navigation Using A Smartphone, Yaohua Chang

Dissertations and Theses

We propose an integrated solution of indoor navigation using a smartphone, especially for assisting people with special needs, such as the blind and visually impaired (BVI) individuals. The system consists of three components: hybrid modeling, real-time navigation, and client-server architecture. In the hybrid modeling component, the hybrid model of a building is created region by region and is organized in a graph structure with nodes as destinations and landmarks, and edges as traversal paths between nodes. A Wi-Fi/cellular-data connectivity map, a beacon signal strength map, a 3D visual model (with destinations and landmarks annotated) are collected while a modeler walks …

V-Slam And Sensor Fusion For Ground Robots, Ejup Hoxha

Dissertations and Theses

In underground, underwater and indoor environments, a robot has to rely solely on its on-board sensors to sense and understand its surroundings. This is the main reason why SLAM gained the popularity it has today. In recent years, we have seen excellent improvement on accuracy of localization using cameras and combinations of different sensors, especially camera-IMU (VIO) fusion. Incorporating more sensors leads to improvement of accuracy,but also robustness of SLAM. However, while testing SLAM in our ground robots, we have seen a decrease in performance quality when using the same algorithms on flying vehicles.We have an additional sensor for ground …

Towards Improving Accuracy And Interpretability Of Deep Learning Based On Satellite Image Classification, Yamile Patino Vargas

Dissertations and Theses

ABSTRACT

The study of satellite images provides a way to monitor changes in the surface of the Earth and the atmosphere. Convolutional Neural Networks (CNN) have shown accurate results in solving practical problems in multiple fields. Some of the more recognized fields using CNNs are satellite imagery processing, medicine, communication, transportation, and computer vision. Despite the success of CNNs, there remains a need to explain the network predictions further and understand what the network is determining as valuable information.

There are several frameworks and methodologies developed to explain how CNNs predict outputs and what their internal representations are [1, 4, …

2d Vector Map And Database Design For Indoor Assisted Navigation, Luciano Caraciolo Albuquerque

Dissertations and Theses

In this paper we implemented a 2D Vector Map, map editor and Database design intended to provide an efficient way to convert cad files from indoor environments to a set of vectors representing hallways, doors, exits, elevators, and other entities embedded in a floor plan, and save them in a database for use by other applications, such as assisted navigation for blind people.

A graphical application as developed in C++ to allow the user to input a CAD DXF file, process the file to automatically obtain nodes and edges, and save the nodes and edges to a database for posterior …

Vehicle Engine Classification Using Of Laser Vibrometry Feature Extraction, Chi Him Liu

Dissertations and Theses

Used as a non-invasive and remote sensor, the laser Doppler vibrometer (LDV) has been used in many different applications, such as inspection of aircrafts, bridge and structure and remote voice acquisition. However, using LDV as a vehicle surveillance device has not been feasible due to the lack of systematic investigations on its behavioral properties. In this thesis, the LDV data from different vehicles are examined and features are extracted. A tone-pitch indexing (TPI) scheme is developed to classify different vehicles by exploiting the engine’s periodic vibrations that are transferred throughout the vehicle’s body. Using the TPI with a two-layer feed-forward …

Evaluating Distributed Word Representations For Predicting Missing Words In Sentences, Saniya Saifee

Dissertations and Theses

In recent years, the distributed representation of words in vector space or word embeddings have become very popular as they have shown significant improvements in many statistical natural language processing (NLP) tasks as compared to traditional language models like Ngram. In this thesis, we explored various state-of-the-art methods like Latent Semantic Analysis, word2vec, and GloVe to learn the distributed representation of words. Their performance was compared based on the accuracy achieved when tasked with selecting the right missing word in the sentence, given five possible options. For this NLP task we trained each of these methods using a training corpus …

Technetium: Productivity Tracking For Version Control Systems, David Leonard

Dissertations and Theses

In recent years, the City College of New York has seen its Computer Science program grow immensely, to the point of overcrowding. This has negative implications for both students and professors, particularly in introductory computer science courses in which constant feedback, iteration and collaboration with others is key to success. In this paper we propose various models for collaboration among students in all course levels using distributed version control systems and implement a secure and efficient tool for visualizing collaborative efforts by observing past work [5]. Lastly, we lay the foundation for future work around additional collaborative metrics, features and …

An Approach To Automatic Detection Of Suspicious Individuals In A Crowd, Satabdi Mukherjee

Dissertations and Theses

This paper describes an approach to identify individuals with suspicious objects in a crowd. It is based on a well-known image retrieval problem as applied to mobile visual search. In many cases, the process of building a hierarchical tree uses k-means clustering followed by geometric verification. However, the number of clusters is not known in advance, and sometimes it is randomly generated. This may lead to a congested clustering which can cause problems in grouping large real-time data. To overcome this problem we have applied the Indian Buffet stochastic process approach in this paper to the clustering problem. We present …

Assessing Satellite Image Data Fusion With Information Theory Metrics, James Cross

Dissertations and Theses

A common problem in remote sensing is estimating an image with high spatial and high spectral resolution given separate sources of measurements from satellite instruments, one having each of these desirable properties. This thesis presents a survey of seven families of algorithms which have been developed to provide this common pattern of satellite image data fusion. They are all tested on artificially degraded sets of satellite data from the Moderate Resolution Imaging Spectroradiometer (“MODIS”) with known ideal results, and evaluated using the commonly accepted data fusion assessment metrics spectral angle mapper (“SAM”) and Erreur Relative Globale Adimensionelle de Synth`ese (“ERGAS”). …

Neuroevolution And An Application Of An Agent Based Model For Financial Market, Anil Yaman

Dissertations and Theses

Market prediction is one of the most difficult problems for the machine learning community. Even though, successful trading strategies can be found for the training data using various optimization methods, these strategies usually do not perform well on the test data as expected. Therefore, selection of the correct strategy becomes problematic. In this study, we propose an evolutionary algorithm that produces a variation of trader agents ensuring that the trading strategies they use are different. We discuss that because the selection of the correct strategy is difficult, a variety of agents can be used simultaneously in order to reduce risk. …

Physical Unclonable Function Techniques Applied For Digital Hardware Protection, Anthony Barrera

Dissertations and Theses

"Privacy is an important property that is growing harder to keep as people develop new ways to steal information from users on their computers. Software alone cannot ensure privacy since an infected system is untrustworthy. This paper presents several challenges malware brings that can be solved by using an external processor. Techniques such as keystroke encryption and message authentication can be used to protect users from having their passwords and other private data stolen. To take advantage of the external hardware, a physical unclonable function can be used to generate private keys without the need for storing them in memory. …

Discrete Transforms With Good Time-Frequency And Spatial-Frequency Localization, David Chisholm

Dissertations and Theses

Discrete orthonormal time-frequency basis functions are described and used for both analysis and synthesis of complex-valued signals. We derive expressions for complex-valued expansion coefficients in time-frequency lattices in the discrete one dimensional case. This derivation is based on Professor I. Gertner's previous construction of a complete orthonormal set of basis functions well localized in the temporal-spatial-frequency domain in the continuous case. We describe how these can be generalized to any number of dimensions. Example applications are presented in one and two dimensions. Three dimensional basis functions are visualized and discussed. Finally, a full a Matlab implementation of this work is …

3d Hallway Modeling Using A Single Image, Gregory M. Olmschenk

Dissertations and Theses

"Real-time, low-resource corridor reconstruction using a single consumer grade RGB camera is a powerful tool for allowing a fast, inexpensive solution to indoor mobility of a visually impaired person or a robot. The perspective and known geometry of a corridor is used to extract the important features of the image and create a 3D model from a single image. Multiple 3D models can be combined to increase confidence and provide a global 3D model. This paper presents our results on 3D corridor modeling using single images. First a simple but effective 3D corridor modeling approach is introduced which makes very …

Re-Engineering Ccny's Business Process Of The Non-Tax Levy Disbursements Requisitions, Ferguson Rodley

Dissertations and Theses

No abstract provided.

Rapid Decoding Of Digital Data Streams Using Field Programmable Gate Arrays, Andrew Hernandez

Dissertations and Theses

No abstract provided.

Developing A Comprehensive Software Toolkit For Creating Digital Mosaic Artwork, Dmitry Bosikov

Dissertations and Theses

No abstract provided.

Parallel Trie-Based Frequent Itemset Mining On Graphics Processors, Jay Junjie Yao

Dissertations and Theses

No abstract provided.

Context Awareness And Discovery For Helping The Blind, Martin Goldberg

Dissertations and Theses

No abstract provided.

Multimedia Content-Based Indexing And Recognition In Digital Libraries, Steven Medina

Dissertations and Theses

"Recognition of digital media is more prevalent in today’s computer culture than ever before. The advent of low cost storage has created a seemingly infinite amount of metadata on the internet, as well as on local machines throughout the world. It is more important than ever to have the capability of quickly and accurately filtering this metadata to find a desired result. Data can be searched using various criteria. For example, text data is searched by analyzing the contents of the text itself. One might execute a search using a method as simple as looking for the ASCII file name, …

Complexity Of Minimum Corridor Guarding Problems, Ning Xu

Dissertations and Theses

"In this paper, the complexity of minimum corridor guarding problems is discussed. These problem can be described as: given a connected orthogo-nal arrangement of vertical and horizontal line segments and a guard with unlimited visibility along a line segment, find a tree or a closed tour with minimum total length along edges of the arrangement, such that if the guard runs on the tree or on the closed tour, all line segments are visited by the guard. These problems are proved to be NP-complete. Keywords: computational complexity, computational geometry, corridor guarding, NP-complete"

Parallel Computing With Improved Techniques For Monte Carlo Simulation In Var, Ping Hung Wu

Dissertations and Theses

"Value at Risk ( VaR ) is a widely used tool for the assessment of one’s investments. VaR is used to evaluate the risk of loss on a financial portfolio. This metric can be computed in several ways. In the historical approach, past trends of the appropriate combination of stocks is used to estimate current portfolio fluctuations. The variance – covariance method, meanwhile, seeks to discover relationships in price fluctuations for one’s stocks. Finally, Monte Carlo simulation capitalizes upon the stochastic nature of stock prices to predict future value. This latter approach, however, relies heavily on the multiplication of vectors …

Acceleration Of Monte Carlo Value At Risk Estimation Using Graphics Processing Unit (Gpu), Wei Wu

Dissertations and Theses

"Value at Risk (VaR) is one of the most popular tools used to estimate the exposure to market risks, and it measures the worst expected loss at a given confidence level. Monte Carlo simulation is one of the best methods to calculate VaR and it is widely used in financial industry. Unfortunately, it is time consuming especially when the simulated samples and the number of assets in a portfolio are very large. The graphics processing unit (GPU) is a specialized multiprocessor which has highly parallel structure supporting more effective than general-purpose CPUs for a range of complex algorithms. In this …