Header menu link for other important links
X
Relative parts: Distinctive parts for learning relative attributes
R.N. Sandeep, , C.V. Jawahar
Published in IEEE Computer Society
2014
Pages: 3614 - 3621
Abstract
The notion of relative attributes as introduced by Parikh and Grauman (ICCV, 2011) provides an appealing way of comparing two images based on their visual properties (or attributes) such as 'smiling' for face images, 'naturalness' for outdoor images, etc. For learning such attributes, a Ranking SVM based formulation was proposed that uses globally represented pairs of annotated images. In this paper, we extend this idea towards learning relative attributes using local parts that are shared across categories. First, instead of using a global representation, we introduce a part-based representation combining a pair of images that specifically compares corresponding parts. Then, with each part we associate a locally adaptive 'significance-coefficient' that represents its discriminative ability with respect to a particular attribute. For each attribute, the significance-coefficients are learned simultaneously with a max-margin ranking model in an iterative manner. Compared to the baseline method, the new method is shown to achieve significant improvement in relative attribute prediction accuracy. Additionally, it is also shown to improve relative feedback based interactive image search. © 2014 IEEE.
CONTINUE READING
Figures & Tables (11)
  • Figure-0
    Figure 1. Given ordered pair of images, first we detect parts ... Expand
  • Figure-1
    Figure 2. Given an input image (left), the parts that ... Expand
  • Figure-2
    Figure 4. Example pairs and their ground-truth annotations from ... Expand
  • Figure-3
    Figure 5. Example pairs from LFW-10 dataset. The images exhibit ... Expand
  • Figure-4
    Figure 3. Input image (left), parts detected using [35] ... Expand
  • Figure-5
    Table 1. Results on PubFig-29 dataset. Though all the methods ... Expand
  • Figure-6
    Table 2. Average relative attribute prediction accuracies using ... Expand
  • Figure-7
    Figure 6. For three attributes from LFW-10 dataset (“smiling”, ... Expand
  • 3 figures hidden
    Show all figures
About the journal
JournalData powered by SciSpaceProceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition
PublisherData powered by SciSpaceIEEE Computer Society
ISSN10636919