Japanese / English

Detail of Publication

Text Language Japanese
Authors Kazuto NOGUCHI, Koichi KISE, Masakazu IWAMURA
Title Efficient Recognition of Specific Objects by Cascading Approximate Nearest Neighbor Searchers
Journal Trans. IEICE
Vol. J92-D
No. 12
Pages pp.2238-2248
Reviewed or not Reviewed
Month & Year December 2009
Abstract This paper concerns the task of recognizing object instances, or specific object recognition, with the special focus on planar object recognition. For this task, it is common to employ local features such as SIFT for indexing images of objects and match them using approximate nearest neighbor search for efficiency. In this paper, we propose a new method of more efficient recognition by taking into account the fact that the appropriate level of accuracy of nearest neighbor search depends on images to be recognized. The proposed method is characterized by the mechanism called cascaded recognizers: multiple recognizers with approximate nearest neighbor search are cascaded so as to determine automatically the appropriate level of approximation depending on the input image. From experimental results with 10,000 images, we have confirmed that the proposed method is capable of achieving a recognition rate of 98% in 1 ms / query, which is 1/10 of the recognition time without the cascade, and 1/40 of the recognition time with conventional approximate nearest neighbor search such as ANN and LSH. In addition, a recognition error rate of the proposed method has been suppressed to 0% by allowing a rejection rate of 8.6%. Experimental results with 100,000 images show high scalability of the proposed method.
URL http://search.ieice.org/bin/summary.php?id=j92-d_12_2238
Back to list