Publications
Detailed Information
A Neural Contextual Model for Detection and Recognition of Text Embedded in Online Images : 온라인 영상의 텍스트 검출 및 인식을 위한 신경망 문맥 모델
DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | 유석인 | - |
dc.contributor.author | 강철무 | - |
dc.date.accessioned | 2017-07-13T07:19:42Z | - |
dc.date.available | 2017-07-13T07:19:42Z | - |
dc.date.issued | 2017-02 | - |
dc.identifier.other | 000000141220 | - |
dc.identifier.uri | https://hdl.handle.net/10371/119258 | - |
dc.description | 학위논문 (박사)-- 서울대학교 대학원 : 전기·컴퓨터공학부, 2017. 2. 유석인. | - |
dc.description.abstract | We address the problem of detecting and recognizing the text embedded in online images that are circulated over the Web. Our idea is to leverage context information for both text detection and recognition. For detection, we use local
image context around the text region, based on that the text often sequentially appear in online images. For recognition, we exploit the metadata associated with the input online image, including tags, comments, and title, which are used as a topic prior for the word candidates in the image. To infuse such two sets of context information, we propose a contextual text spotting network (CTSN).We perform comparative evaluation with ve state-of-the-art text spotting methods on newly collected Instagram and Flickr datasets. We show that our approach that benets from context information is more successful for text spotting in online images. | - |
dc.description.tableofcontents | Chapter 1 Introduction 1
Chapter 2 RelatedWork 5 Chapter 3 Preliminary of Neural Networks 9 3.1 Basic of Neural Network 9 3.2 Convolutional Neural Network 12 3.3 Pooling Layer 13 3.4 Activation Function 15 3.5 Recurrent Neural Network 16 3.6 Back-Propagation Through Time 16 3.7 Bidirectional Recurrent Neural Networks 17 3.8 Long-Short Term Memory 17 3.9 Optimization 18 3.10 Training Loss 19 3.11 Training Process 20 Chapter 4 Approach for Contextual Text Spotting 25 4.1 Overview of the Proposed Framework 25 4.2 Context-Aware Text Detection 26 4.2.1 Text Proposals 27 4.2.2 Text Detection Network 28 4.2.3 Extraction of Textline Boxes 33 4.2.4 Text Detection Network Variants 37 4.3 Context-Aware Word Recognition 40 4.3.1 Bias Networks for Context 42 4.3.2 Recurrent Word Recognition Network 43 4.3.3 Recognition Network Variant 46 Chapter 5 Experiments 52 5.1 Dataset 52 5.2 Experimental Setup 55 5.3 Training 56 5.4 Hyperparameters of Contextual Model 59 5.5 Neural Network Architecture Variants 71 5.6 Results 74 Chapter 6 Conclusion 86 요약 102 | - |
dc.format | application/pdf | - |
dc.format.extent | 11159963 bytes | - |
dc.format.medium | application/pdf | - |
dc.language.iso | en | - |
dc.publisher | 서울대학교 대학원 | - |
dc.subject | Text Detection | - |
dc.subject | Text Recognition | - |
dc.subject | Context Model | - |
dc.subject | Deep learning | - |
dc.subject.ddc | 621 | - |
dc.title | A Neural Contextual Model for Detection and Recognition of Text Embedded in Online Images | - |
dc.title.alternative | 온라인 영상의 텍스트 검출 및 인식을 위한 신경망 문맥 모델 | - |
dc.type | Thesis | - |
dc.description.degree | Doctor | - |
dc.citation.pages | xiii, 102 | - |
dc.contributor.affiliation | 공과대학 전기·컴퓨터공학부 | - |
dc.date.awarded | 2017-02 | - |
- Appears in Collections:
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.