Author | Conference | Journal | Organization | Year | DOI
Look for results that meet for the following criteria:
Academic
Publications
Real-Time Visual Concept Classification

Real-Time Visual Concept Classification,10.1109/TMM.2010.2052027,IEEE Transactions on Multimedia,Jasper R. R. Uijlings,Arnold W. M. Smeulders,Remko J.

Edit
Real-Time Visual Concept Classification   (Citations: 2)
BibTex | RIS | RefWorks Download
As datasets grow increasingly large in content-based image and video retrieval, computational efficiency of concept classification is important. This paper reviews techniques to accelerate concept classification, where we show the trade-off between computational efficiency and accuracy. As a basis, we use the Bag-of-Words algorithm that in the 2008 benchmarks of TRECVID and PASCAL lead to the best performance scores. We divide the evaluation in three steps: 1) Descriptor Extraction, where we evaluate SIFT, SURF, DAISY, and Semantic Textons. 2) Visual Word Assignment, where we compare a k-means visual vocabulary with a Random Forest and evaluate subsampling, dimension reduction with PCA, and division strategies of the Spatial Pyramid. 3) Classification, where we evaluate the χ2, RBF, and Fast Histogram Intersection kernel for the SVM. Apart from the evaluation, we accelerate the calculation of densely sampled SIFT and SURF, accelerate nearest neighbor assignment, and improve accuracy of the Histogram Intersection kernel. We conclude by discussing whether further acceleration of the Bag-of-Words pipeline is possible. Our results lead to a 7-fold speed increase without accuracy loss, and a 70-fold speed increase with 3% accuracy loss. The latter system does classification in real-time, which opens up new applications for automatic concept classification. For example, this system permits five standard desktop PCs to automatically tag for 20 classes all images that are currently uploaded to Flickr.
Journal: IEEE Transactions on Multimedia - TMM , vol. 12, no. 7, pp. 665-681, 2010
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
More »
    • ...Bag-of-Visual-Features(BoVF) has become a popular method for object recognition and visual categorization for its effectiveness and flexibility[1], [2], showing robustness to occlusion and also to several kinds of variations that normally curse object detection methods, such as those derived from lighting conditions, scale, shape and rotations...
    • ...A typical high-level semantic BoVW image classification model which forms the basis of several state-of-the-art image/video retrieval systems [1], [2] is given in figure 1. The model includes three steps: 1) Feature extraction using SURF (4 × 4) algorithm which contains robust color information and global content extraction of local features [1] [6] [10] [11] Color space that has invariant characteristic to illumination especially ...
    • ...A typical high-level semantic BoVW image classification model which forms the basis of several state-of-the-art image/video retrieval systems [1], [2] is given in figure 1. The model includes three steps: 1) Feature extraction using SURF (4 × 4) algorithm which contains robust color information and global content extraction of local features [1] [6] [10] [11] Color space that has invariant characteristic to illumination especially ...
    • ... high-level semantic dictionary by random forest algorithm[2] to extract the visual primitives for classifying low-level visual vocabulary, and re-classification of mid-level vocabulary, spatial relations and the integration of relevant characteristics of the context [11][12][13] And ventually build a high differentiation in high-level semantic visual dictionary; 3) Classification using 2 kernel χ − SVM classification algorithm [1]...

    Lintao Lv, et al. Pornographic images detection using High-Level Semantic features

    • ...[1,2,3]), but differ in that instances and the feature space are defined in ways that exploit the structure of classroom learning and the nature of the assessment system...

    Qifeng Qiao, et al. Classroom Video Assessment and Retrieval via Multiple Instance Learnin...

Sort by:

Citations (2)

Share this on