Date of Award

Spring 5-2020

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Computer Science

First Advisor

Mick O'Neal

Abstract

Deep learning is a subset of machine learning that extracts high-level features from raw data over multiple layers. Deep learning for computer vision has become more popular in the last few years, but the data and resource requirements make it difficult to implement on Internet of Things (IoT) devices. This work aims at providing a series of techniques that can alleviate some of the strain caused by these requirements in order to make a computer vision system for inventory management more feasible. These techniques include data collection, data preprocessing, transfer learning, and a method for intelligently growing a dataset throughout the lifetime of the system. The techniques laid out in this thesis are a combination of existing and proposed techniques to aid in the use of a computer vision system throughout its lifecycle.

Download

Share

COinS