(Downloads - 0)
For more info about our services contact : help@bestpfe.com
Table of contents
1 Introduction
1.1 Background
1.2 Motivation and aim of study
1.2.1 Building a category-level object detector
1.2.2 Aligning to build a similarity measure
1.2.3 A direct approach to the misalignment problem
1.2.4 Aligning rigid objects
1.2.5 Aligning deformable objects
1.3 Graph matching
1.4 Objective and contributions of the thesis
2 A tensorial formulation for hyper-graph matching
2.1 Introduction
2.1.1 Motivation and goals
2.1.2 Problem statement
2.1.3 Organization and Contributions
2.2 Previous work
2.2.1 Historical review of literature
2.2.2 Spectral matching
2.2.3 Power iterations for eigenvalue problems
2.3 Proposed approach
2.3.1 Tensor formulation of hypergraph matching
2.3.2 Tensor power iterations
2.3.3 Tensor power iterations for unit-norm rows
2.3.4 Merging potentials of different orders
2.3.5 `1-norm constraint for rows
2.3.6 Building tensors for computer vision
2.4 Implementation
2.4.1 Separable similarity measure
2.5 Experiments
2.5.1 Synthetic data
2.5.2 House dataset
2.5.3 Natural images
2.5.4 3D object matching
2.5.5 Potentials of different orders
2.6 Conclusion
3 Graph matching for image categorization
3.1 Introduction
3.1.1 Motivation and goals
3.1.2 Organization and contributions
3.2 Previous work
3.2.1 Review of literature
3.2.2 Caputo’s method
3.3 Proposed approach
3.3.1 Image representation
3.3.2 Matching two images
3.3.3 A kernel for image comparison
3.4 Implementation
3.4.1 Ishikawa’s method
3.4.2 Proposed method: Curve expansion
3.5 Experiments
3.5.1 Running time
3.5.2 Image matching
3.5.3 Image classification
3.6 Conclusion
4 Image alignment for object detection
4.1 Introduction
4.1.1 Motivation and goals
4.2 Proposed approach
4.2.1 Image Model
4.2.2 Efficient Similarity Computation
4.3 Proof of concept: automated object discovery
4.4 Learning a detector
4.4.1 Latent SVMs
4.4.2 Hybrid method: Latent SVM and exemplar SVM
4.5 Implementation and Results
4.6 Conclusion
5 Conclusion and perspectives
5.1 Contributions
5.2 Future work
5.2.1 Detection experiments
5.2.2 Aspects and full object model
5.2.3 Joint Alignment of multiple images
5.3 Other work
A Power iterations for unit norm rows
