Deep convolutional neural networks with an application towards geospatial object recognition /

Batı, Emrecan
The passion of human-being to invent intelligent systems becomes more and more meaningful day by day, as the data captured every second by artificial sensors needs to be examined and classified for many applications. The processing of ever-increasing amount of data by defining information explicitly seems nearly impossible, regarding the variability and the amount of the information, which reveals the need for intelligent systems that are capable of learning. Deep learning is a set of algorithms that attempts to find a hierarchical representation of the input data by trying to mimic the way human brain captures the critical aspects of excessive sensory data, to which it is exposed to every second. Convolutional neural networks, which are trainable learning structures, are also biologically inspired from the receptive fields in visual cortex. In this thesis, the performance of convolutional neural networks are investigated for an application towards geospatial target detection and classification from satellite images. Based on the experiments, it is observed that the utilization of preprocessing, dropout, i.e. dropping neurons randomly in the training phase, and rectified linear unit as the activation function improves the classification rate, significantly. However, the application of this deep classifier on satellite images still yields high false alarm rate, possibly due to insufficient number of training data.


Training Methodology for a Multiplication Free Implementable Operator Based Neural Networks
Yıldız, Ozan; Yarman Vural, Fatoş Tunay; Department of Computer Engineering (2017)
Technological advances opened new possibilities for computing environments including smart phones, smart appliances, and drones. Engineers try to make these devices smart, self-sustaining through usage of machine learning techniques. However, most of the mobile environments have limited resources like memory, computing power and battery, and consequently traditional machine learning algorithms which require relatively high resources might not be suitable for them. Therefore, efficient versions of traditiona...
Semantic data modeling of spatiotemporal database applications
Yazıcı, Adnan; Sun, N (Wiley, 2001-07-01)
Due to the ubiquity of space-related and time-related information, the ability of a database system to deal with both spatial and temporal phenomenon facts in a spatiotemporal applications is highly desired. However, uncertain and fuzzy information in these applications highly increases the complexity of database modeling. In this paper we introduce a semantic data modeling approach for spatiotemporal database applications. We specifically focus on various aspects of spatial and temporal database issues and...
Adaptive mean-shift for automated multi object tracking
Beyan, C.; Temizel, Alptekin (2012-01-01)
Mean-shift tracking plays an important role in computer vision applications because of its robustness, ease of implementation and computational efficiency. In this study, a fully automatic multiple-object tracker based on mean-shift algorithm is presented. Foreground is extracted using a mixture of Gaussian followed by shadow and noise removal to initialise the object trackers and also used as a kernel mask to make the system more efficient by decreasing the search area and the number of iterations to conve...
Online mining of human deep intention by proactive environment changes using deep neural networks
Er, Nur Baki; Erkmen, Aydan Müşerref; Department of Electrical and Electronics Engineering (2015)
This thesis focuses on surfacing human deep intention, which is known or assumed, in a smart environment that consists of autonomous robotic systems which can interact with the human. Deep intentions are defined as kind of actions that humans would like to behave but pushed deeper in the stack of the intentions in a daily life. The purpose of the designed system is to observe the human in the smart room for a while and to analyze human’s behaviors to offer the optimal set of system behavior to surface a des...
Supervised Learning in Football Game Environments Using Artificial Neural Networks
Baykal, Ömer; Alpaslan, Ferda Nur (2018-09-23)
Game industry has become one of the sectors that commonly use artificial intelligence. Today, most of the game environments include artificial intelligence agents to offer more challenging and entertaining gameplay experience. Since it gets harder to develop good agents as games become more complex, machine learning methods have started to be used in some notable games to shorten the development process of agents and to improve their quality. Popularity of machine learning applications in game environments ...
Citation Formats
E. Batı, “Deep convolutional neural networks with an application towards geospatial object recognition /,” M.S. - Master of Science, Middle East Technical University, 2014.