Date of Award
2019
Document Type
Open Access Master's Report
Degree Name
Master of Science in Computer Science (MS)
Administrative Home Department
Department of Computer Science
Advisor 1
Nilufer Onder
Advisor 2
Nina Mahmoudian
Committee Member 1
Bo Chen
Committee Member 2
Jianhui Yue
Abstract
Object detection and classification plays a major role in today's modern technology. The implementations of these concepts range from consumer products to self driving cars. These concepts largely reply on the data sets used for training these models. There is a considerable amount of effort in generating these data sets for every specific application of these algorithms.
In this report, a method for generating image data sets with the use of visual feature tracking and deep learning algorithms for application in autonomous vehicles has been proposed. The aim is to reduce the time and effort dedicated towards the generation of these application specific data sets.
For this purpose, a software has been developed in Python for a Linux based system using Tensorflow, Keras, Pygames and OpenCV libraries which is capable of tracking an object of interest in a given media input specified by the user along with detecting various similar objects using a pre-trained Classification neural network. This software then compiles a file containing all the annotations for the above specified objects.
Recommended Citation
Pallapotu, Kusuma, "DATA SET GENERATION USING DEEP LEARNING ALGORITHMS AND VISUAL FEATURE TRACKING", Open Access Master's Report, Michigan Technological University, 2019.