Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
CVPR '06 Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2
| Hi-index | 0.00 |
This paper develops a supervised nonlinear subspace of bag-of-features for category classification. Bag-of-features represents an image as an orderless distribution of features, which selects the visual words by clustering and uses the similarity with each visual word as the features for classification. In this paper, we propose to model the ensemble of visual words with a supervised nonlinear neighborhood embedding method to a more discriminative space for category classification. The supervised nonlinear neighborhood embedding(SNNE) is used to model visual words and extract the discrimitive information specialized for each category. The projection length in subspace is used as features for classification. The SNNE subspace method can model the nonlinear variations induced by various kinds of visual words and extract more discriminative feature for object recognition. The proposed method is evaluated using the Cal-tech and GRAZ01 database. We confirm that the proposed method is comparable with state-of-the-art methods without absolute position information.