Curved text detection and ground truth generation for natural scene images / Ch’ng Chee Kheng

Ch’ng, Chee Kheng (2018) Curved text detection and ground truth generation for natural scene images / Ch’ng Chee Kheng. Masters thesis, University of Malaya.

[img] PDF (The Candidate's Agreement)
Restricted to Repository staff only

Download (201Kb) | Request a copy
    [img]
    Preview
    PDF (Thesis M.A)
    Download (3797Kb) | Preview

      Abstract

      At present, text orientation is not diverse enough in the existing scene text datasets. For instance, text with curve-orientation has close to zero existence in them and thus received minimal attention from the community. Motivated by this phenomenon, a new scene text dataset, Total-Text, which emphasized on text orientations diversity has been collected as the major contribution of this work. It is the first properly scaled scene dataset that features three different text orientations: horizontal, multi-oriented, and curve-oriented. In addition, several studies regarding other important elements such as the practicality and quality of groundtruth, evaluation protocol, insights of curved text, and the annotation process are presented in this work as well. These elements are found to be as important as the images and groundtruth to facilitate a new research direction. In addition, Polygon- Faster-RCNN, a text detection baseline, has been proposed as the second major contribution of this work. It has demonstrated its ability in detecting text in all kinds of orientations. Images of Total-Text and its annotation are available at https://github.com/cschan/ Total-Text-Dataset.

      Item Type: Thesis (Masters)
      Additional Information: Dissertation (M.A.) – Faculty of Computer Science & Information Technology, University of Malaya, 2018.
      Uncontrolled Keywords: Curved text; Scene text detection; Natural scene images; Dataset; Annotation process
      Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
      Divisions: Faculty of Computer Science & Information Technology
      Depositing User: Mr Mohd Safri Tahir
      Date Deposited: 07 Jan 2020 06:35
      Last Modified: 18 Jan 2020 10:36
      URI: http://studentsrepo.um.edu.my/id/eprint/10702

      Actions (For repository staff only : Login required)

      View Item