Handbook of Deep Learning Applications

Abstract

6.3 OCR for Latin Script Intro; Contents; Designing a Neural Network from Scratch for Big Data Powered by Multi-node GPUs; 1 Introduction; 2 A Primer on Neural Networks; 3 A Mathematical Formalization of Neural Networks; 4 Problem and Dataset; 5 A Neural Network in Python; 6 A Distributed Neural Network Using a Message Queue for Communication; 7 A GPU-Powered Neural Network; 8 Discussion and Homework; 9 Conclusion; References; Deep Learning for Scene Understanding; 1 Introduction; 2 Object Recognition; 2.1 Object Recognition Pipeline; 2.2 Hand-Crafted Features for Object Recognition 2.3 Deep Learning Techniques for Object Recognition3 Face Detection and Recognition; 3.1 Non-deep Learning Techniques for Face Detection and Recognition; 3.2 Deep Learning for Face Detection and Recognition; 4 Text Detection in Natural Scenes; 4.1 Classical Approaches for Text Detection; 4.2 Deep Networks for Text Detection; 5 Depth Map Estimation; 5.1 Methodology of Depth Map Estimation; 5.2 Depth Map Estimation Using Pattern Matching; 5.3 Deep Learning Networks for Depth Map Estimation; 6 Scene Classification; 6.1 Scene Classification Using Handcrafted-Features 6.2 Scene Classification Using Deep Features7 Caption Generation; 7.1 Deep Networks for Caption Generation; 8 Visual Question Answering (VQA); 8.1 Deep Learning Methods for VQA; 9 Integration of Scene Understanding Components; 9.1 Non-deep Learning Works for Holistic Scene Understanding; 9.2 Deep Learning Based Works for Holistic Scene Understanding; 10 Conclusion; References; An Application of Deep Learning in Character Recognition: An Overview; 1 Introduction; 2 Objectives of Document Analysis 2.1 Extraction of Properties (Metadata) for Indexing and for the Provision of Filter Criteria for the Search2.2 Classification of Documents Based on Specific Categories; 2.3 Automatic Creation of Company-Specific Dictionaries; 2.4 Statistics on Various Properties of Document Contents; 2.5 Automatic Translation; 3 Application of the Automated Document Analysis; 3.1 Historic Document Analysis; 3.2 Document Layout Analysis; 3.3 Text Extraction Form Scanned Documents and Digitizing the Information; 3.4 Automated Traffic Monitoring, Surveillance and Security Systems 3.5 Automated Postal-Mail Sorting4 Significance of Deep Learning over Machine Learning; 4.1 Deep Learning Techniques and Architecture; 5 Peculiarities and Challenges for OCR with Deep Learning; 5.1 Dataset; 5.2 Data Interoperability and Data Standards; 5.3 Build and Integrate Big Image Dataset; 5.4 Language and Script Peculiarities; 5.5 Black Box and Deep Learning; 5.6 Processing Hardware Power; 5.7 Implementation (Available Libraries) Can Be Hardware Dependent; 6 Machine Learning, Deep Learning and Optical Character Recognition; 6.1 OCR for Arabic like Script; 6.2 OCR for Symbolic Script