Unsupervised Object Discovery from Images by Mining Local Features Using Hashing
In this paper, we propose a new methodology for efficiently discovering objects from images without supervision. The basic idea is to search for frequent patterns of closely located features in a set of images and consider a frequent pattern as a meaningful object class. We develop a system for discovering objects from segmented images. This system is implemented by hashing only. We present experimental results to demonstrate the robustness and applicability of our approach.
KeywordsHash Function Frequent Pattern Object Class Mining Local Locate Component
- 5.Indyk, P., Motwani, R.: Approximate nearest neighbors: Towards removing the curse of dimensionality. In: Thirtieth Annual ACM Symposium on the Theory of Computing, pp. 604–613 (1998)Google Scholar
- 6.Haveliwala, T.H., Gionis, A., Indyk, P.: Scalable techniques for clustering the web. In: Third International Workshop on the Web and Databases, pp. 129–134 (2000)Google Scholar
- 7.Tsunoda, N., Watanabe, T., Sugawara, K.: Image segmentation by adaptive thresholding of minimum spanning trees. Transactions of the Institute of Electronics, Information and Communication Engineers J87-D-II(2), 586–594 (2004)Google Scholar