The goal of this project is to take content-based image-retrieval one step further in size and closer to real world applications. The system handles over 10 Million images to date, and the collection is still growing. Cortina demo got updated up to version 3.
Recent advances in processing and networking capabilities of computers have led to an accumulation of immense amounts of multimedia data such as images. One of the largest repositories for such data is the World Wide Web (WWW). We present Cortina, a large-scale image retrieval system for the WWW. Cortina v.3 indexes still over 10 Million images using image content, text and annotations At the systems level, the components of Cortina include building image collections using a Web crawler, collecting category information and keywords, and processing images to compute content descriptors. In this last version of Cortina the user has 4 options to start a search as shown in the Screen Shot on the right. Keyword query, to do a keyword or text search within the existing images, upload an image or insert the URL, browse images in the database randomly, or cluster to visualize images in its semantic clusters.
The new functionalities of Cortina include duplicate image detection, category and image content based search, face detection and relevance feedback.
Elisa Drelie Gelasca, Zhiqiang Bi, Steffen Gauglitz, Pratim Ghosh, Emily Moxley, Joriz De Guzman, JieJun Xu, Amir M. Rahimi and B.S.Manjunath
» Semi-automatic image segmentation & annotation tool: This is a set of MATLAB scripts for semi-automatic image segmentation & annotation. It uses a segmentation algorithm to preprocess the image and allows the user to recombine, refine and create segments and to label them out of a set of given categories. more...
For the annotation tool:
B. C. Russell, A. Torralba, K. P. Murphy, W. T. Freeman, LabelMe: a database and web-based tool for image annotation. MIT AI Lab Memo AIM-2005-025, September, 2005.
It handles over 10 Million images to date. The system retrieves images based on visual features and collateral text. We show that a search process which consists of an initial query-by-keyword or query-by-image and followed by relevance feedback on the visual appearance of the results is possible for large-scale data sets. We also show that it is superior to the pure text retrieval commonly used in large-scale systems. Semantic relationships in the data are explored and exploited by data mining, and multiple feature spaces are included in the search process.
B. Sumengen, M. Hol, F.E. Kalsbeek and B.S.Manjunath
Recent advances in processing and networking capabilities of computers have led to an accumulation of immense amounts of multimedia data such as images. One of the largest repositories for such data is the World Wide Web (WWW). We present Cortina, a large-scale image retrieval system for the WWW. This version handled over 3 Million images to date. The system retrieves images wasbased on visual features and collateral text. Methods were introduced to investigate these multi-modal characteristics of the data and to gain insights into the semantics within the data.
Till Quack, Lars Thiele and B.S.Manjunath
Abstract preview: "We present an image search and retrieval system, Cortina, that indexes over 10 Million images using image content, text and annotations. This large collection of image data, gathered from the World Wi..." [more]
Abstract preview: "In this paper we propose a novel framework to obtain a very compact image signature (32 bits) which is invariant to rotation, translation, scaling and other minor perturbations like smoothing, random ..." [more]
Abstract preview: "Recent advances in processing and networking capabilities of computers have led to an accumulation of immense amounts of multimedia data such as images. One of the largest repositories for such data i..." [more]
Abstract preview: "Recent advances in processing and networking capabilities of computers have led to an accumulation of immense amounts of multimedia data such as images. One of the largest repositories for such data i..." [more]
© 2000-2006 Vision Research Lab, UCSB
Last updated:
Credits & Contact
Page visits:
View Stats