Object retrieval with large vocabularies and fast spatial matching Export

@inproceedings{Philbin:2007:VocabTreeAKM, abstract = {In this paper, we present a large-scale object retrieval system. The user supplies a query object by selecting a region of a query image, and the system returns a ranked list of images that contain the same object, retrieved from a large corpus. We demonstrate the scalability and performance of our system on a dataset of over 1 million images crawled from the photo-sharing site, Flickr [3], using Oxford landmarks as queries. Building an image-feature vocabulary is a major time and performance bottleneck, due to the size of our dataset. To address this problem we compare different scalable methods for building a vocabulary and introduce a novel quantization method based on randomized trees which we show outperforms the current state-of-the-art on an extensive ground-truth. Our experiments show that the quantization has a major effect on retrieval quality. To further improve query performance, we add an efficient spatial verification stage to re-rank the results returned from our bag-of-words model and show that this consistently improves search quality, though by less of a margin when the visual vocabulary is large. We view this work as a promising step towards much larger, "web-scale" image corpora.}, author = {Philbin, James and Chum, Ondrej and Isard, Michael and Sivic, Josef and Zisserman, Andrew}, booktitle = {Computer Vision and Pattern Recognition, 2007. CVPR '07. IEEE Conference on}, citeulike-article-id = {1552688}, citeulike-linkout-0 = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4270197}, journal = {Computer Vision and Pattern Recognition, 2007. CVPR '07. IEEE Conference on}, keywords = {approximation, kd-tree, matching, nearest-neighbour, retrieval, thesis, tree, vocabulary}, pages = {1--8}, posted-at = {2007-10-01 23:38:19}, priority = {0}, title = {Object retrieval with large vocabularies and fast spatial matching}, url = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4270197}, year = {2007} }

TY - CONF ID - Philbin:2007:VocabTreeAKM L3 - citeulike-article-id:1552688 N2 - In this paper, we present a large-scale object retrieval system. The user supplies a query object by selecting a region of a query image, and the system returns a ranked list of images that contain the same object, retrieved from a large corpus. We demonstrate the scalability and performance of our system on a dataset of over 1 million images crawled from the photo-sharing site, Flickr [3], using Oxford landmarks as queries. Building an image-feature vocabulary is a major time and performance bottleneck, due to the size of our dataset. To address this problem we compare different scalable methods for building a vocabulary and introduce a novel quantization method based on randomized trees which we show outperforms the current state-of-the-art on an extensive ground-truth. Our experiments show that the quantization has a major effect on retrieval quality. To further improve query performance, we add an efficient spatial verification stage to re-rank the results returned from our bag-of-words model and show that this consistently improves search quality, though by less of a margin when the visual vocabulary is large. We view this work as a promising step towards much larger, "web-scale" image corpora. JF - Computer Vision and Pattern Recognition, 2007. CVPR '07. IEEE Conference on EP - 8 TI - Object retrieval with large vocabularies and fast spatial matching T2 - Computer Vision and Pattern Recognition, 2007. CVPR '07. IEEE Conference on SP - 1 KW - approximation KW - kd-tree KW - matching KW - nearest-neighbour KW - retrieval KW - thesis KW - tree KW - vocabulary AU - Philbin, James AU - Chum, Ondrej AU - Isard, Michael AU - Sivic, Josef AU - Zisserman, Andrew PY - 2007/// UR - http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=4270197 ER -