Learning visual ideals
We address the problem of describing, recognizing, and learning generic, free-form objects in real-world scenes. We introduce a hybrid appearance-based approach, IDEAL, where objects are encoded as a loose collections of parts and the relations between them. The key features of this new approach are the structural part decomposition combining multi-scale wavelet segmentation and hierarchical blobs, and learning to recognize generic object categories, exhibiting large intra-class variability, from real examples with automatic model acquisition.
KeywordsPart Path Segment Segment Blob Feature Evidence Vector Generic Object Category
- 1.W. Bischof and T. Caelli. Learning structural descriptions: a new technique for conditional clustering and rule generation. PR, pages 689–697, 1994.Google Scholar
- 3.M. Burge, W. Burger, and W. Mayr. Recognition and learning with polymorphic structural components. In 13th ICPR, v 1, pp 19–23, Vienna, Austria, 1996.Google Scholar
- 4.W. Burger and B. Bhanu. Signal-to-symbol conversion for structural object recognition using hidden Markov models. In Proc. ARPA Image Understanding Workshop, pages 1287–1291, Monterey, CA, 1994.Google Scholar