Yüksek Lisans Tezleri

Permanent URI for this collectionhttps://hdl.handle.net/20.500.11779/1785

Browse

Filters

20182

Settings

Author: search.filters.author.Gökmen, MuhittinPublisher: search.filters.publisher.MEF Üniversitesi, Fen Bilimleri Enstitüsü

Search Results

Now showing 1 - 2 of 2
  • Master Term Project
    Pre-Ocr Image Optimization by Reinforcement Learning
    (MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Gökmen, Muhittin; Gökmen, Muhittin; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF University
    Optical Character Recognition technology usage in digital transformation of documents is steadily growing by the help of new hardware and software technologies. However digital image optimization for more accurate OCR results continues to be a problem. In this study, we propose a reinforcement learning based model that learns optimal set of actions to increase OCR accuracy in computer screenshot images. Model input images are identified by their grayscale histogram distributions. An unprocessed base image having 100% OCR accuracy is taken initially. The correlation between the grayscale histograms of base image and input image is used for comparison. We implemented reinforcement learning’s random (or optimal) action and reward approach for creating a Q-table. For measuring image to text conversion success, Tesseract OCR software is used. The introduced approach can improve OCR accuracy especially in bulk image to document conversion jobs. By using optimal actions for single image or bulk images, it can also decrease computational load and time-consumption in image processing.
  • Master Term Project
    Building Footprint Extraction Using Deep Learning Techniques
    (MEF Üniversitesi, Fen Bilimleri Enstitüsü, 2018) Gökmen, Muhittin; Gökmen, Muhittin; 02.02. Department of Computer Engineering; 02. Faculty of Engineering; 01. MEF University
    Geospatial industry is getting bigger and bigger these days in addition to creating massive amount of data. Today map features such as roads, building footprints are created through manual techniques. There is not automated solution that extracts map features such as roads, building footprints from satellite imagery. Advance automated feature extraction techniques will serve important uses of map data including disaster response. SpaceNet is a commercial satellite imagery and labeled training data to foster innovation in the development of computer vision algorithms. In this paper we will give a brief explanation about image classification, object recognition processes and why deep learning is effective on object recognition, and how we can apply these concepts to our problem which is Building Footprint extraction. And we will use SpaceNet’s dataset and apply tensorflow backhand object detection model.