Unsupervised Visual Alignment With Similarity Graphs

Alignment of semantically meaningful visual patterns, such as object classes, is an important pre-processing step for a number of applications such as object detection and image categorization. Considering the expensive manpower spent on the annotation of supervised alignment methods, unsupervised alignment techniques are more favourable especially for large-scale problems. Fine adjustment can be effectively and efficiently achieved with the recent image congealing methods, but they require moderately good initialisation which is largely invalid in practice. It remains as an open problem how to align images of class examples with large view point changes. Feature-based methods can solve the problem to some degree, but require manual selection of a good seed image and omit the fact that two examples of semantical classes can be visually very different (e.g., Harley-Davison and Scooter``motorbikes''). In this work, we adopt the feature basedapproach,but to overcome the aforementioned drawbacks define visual similarity as an assignment problem which is solved by fast approximation and non-linear optimization.From pair-wise image similarities we construct an image graph which is used to step-wise align,``morph'', an image to another by graph traveling. Our method also automatically finds a suitable seed by novel centrality measure which identifies ``similarity hubs'' in the graph. The proposed approach in the unsupervised manner outperforms the state-of-the-art methods with classes from the popular benchmark datasets.

PDF Abstract


Results from the Paper

  Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers.


No methods listed for this paper. Add relevant methods here