Speeded Up Robust Features (SURF)

For more info about our services contact : help@bestpfe.com

Table of contents

Abstract
List of Figures
List of Tables
1 Introduction
1.1 Motivation
1.2 Challenges
1.3 Hypotheses
1.4 Contributions
1.5 Outline
2 Literature Review
2.1 Low-level Visual Feature Extraction
2.1.1 Feature Spaces
2.1.2 Local versus Global Features
2.1.3 Local Feature Detection Operators
2.1.4 Feature Description
2.2 Mid-level Image Representations: BoW Models and Extensions
2.2.1 Early Techniques
2.2.2 Current Formalism
2.2.3 BoW-based Approaches
2.3 Feature Normalization
2.3.1 Dimensionality Reduction
2.3.2 BoW Normalization Techniques
2.4 Machine Learning Algorithms for Image Classification
2.4.1 Support Vector Machine
2.4.2 Ensemble Techniques
2.4.3 k-Nearest Neighbor
2.4.4 Visual Codebook Learning
2.5 Other Approaches for Image Classification
2.5.1 Biologically-inspired Models
2.5.2 Deep Models
2.5.3 Part-based Category Models
2.6 Conclusion
3 Challenges and Benchmarks Addressed
3.1 MIRFLICKR Challenge
3.2 ImageCLEF Evaluation Campaign
3.2.1 ImageCLEF 2011 Photo Annotation Challenge
3.2.2 ImageCLEF 2012 Photo Annotation Challenge
3.3 PASCAL VOC Challenge
3.4 15-Scenes Dataset
3.5 Oxford Flowers Dataset
3.6 Conclusion
4 BossaNova Representation
4.1 Coding & Pooling Matrix Representation
4.2 Early Ideas
4.3 BossaNova Pooling Formalism
4.4 Localized Soft-Assignment Coding
4.5 Normalization Strategy
4.6 Computational Complexity
4.7 BossaNova & Fisher Vector: Pooling Complementarity
4.8 BossaNova as a Fisher Kernel Formalism
4.9 Conclusion
5 Experimental Results
5.1 BOSSA to BossaNova Improvements Analysis
5.2 BossaNova Parameter Evaluation
5.2.1 Codebook Size
5.2.2 Bin quantization
5.2.3 Minimum Distance αmin
5.3 Comparison of State-of-the-Art Methods
5.3.1 Results for MIRFLICKR
5.3.2 Results for ImageCLEF 2011 Photo Annotation
5.3.3 Results for PASCAL VOC 2007
5.3.4 Results for 15-Scenes
5.4 BossaNova in the ImageCLEF 2012 Challenge
5.5 Conclusion
6 Application: Pornography Detection
6.1 Related Work
6.2 The Pornography Dataset
6.3 Our Scheme
6.4 Experiments
6.4.1 Experimental Setup
6.4.2 Results
6.4.3 Discussion
6.5 Conclusion
7 Conclusion
7.1 Contributions
7.2 Future Work
7.3 Publications
A BossaNova Fisher Derivation
Bibliography