Relative Attribute Learning with Deep Attentive Cross-image Representation

Zeshang Zhang; Yingming Li; Zhongfei Zhang

Relative Attribute Learning with Deep Attentive Cross-image Representation

Zeshang Zhang, Yingming Li, Zhongfei Zhang ;

Proceedings of The 10th Asian Conference on Machine Learning, PMLR 95:879-892, 2018.

Abstract

In this paper, we study the relative attribute learning problem, which refers to comparing the strengths of a specific attribute between image pairs, with a new perspective of cross-image representation learning. In particular, we introduce a deep attentive cross-image representation learning (DACRL) model, which first extracts single-image representation with one shared subnetwork, and then learns attentive cross-image representation through considering the channel-wise attention of concatenated single-image feature maps. Taking a pair of images as input, DACRL outputs a posterior probability indicating whether the first image in the pair has a stronger presence of attribute than the second image. The whole network is jointly optimized via a unified end-to-end deep learning scheme. Extensive experiments on several benchmark datasets demonstrate the effectiveness of our approach against the state-of-the-art methods.

Related Material

Download PDF

@InProceedings{pmlr-v95-zhang18d, title = {Relative Attribute Learning with Deep Attentive Cross-image Representation}, author = {Zhang, Zeshang and Li, Yingming and Zhang, Zhongfei}, booktitle = {Proceedings of The 10th Asian Conference on Machine Learning}, pages = {879--892}, year = {2018}, editor = {Zhu, Jun and Takeuchi, Ichiro}, volume = {95}, series = {Proceedings of Machine Learning Research}, address = {}, month = {14--16 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v95/zhang18d/zhang18d.pdf}, url = {http://proceedings.mlr.press/v95/zhang18d.html}, abstract = {In this paper, we study the relative attribute learning problem, which refers to comparing the strengths of a specific attribute between image pairs, with a new perspective of cross-image representation learning. In particular, we introduce a deep attentive cross-image representation learning (DACRL) model, which first extracts single-image representation with one shared subnetwork, and then learns attentive cross-image representation through considering the channel-wise attention of concatenated single-image feature maps. Taking a pair of images as input, DACRL outputs a posterior probability indicating whether the first image in the pair has a stronger presence of attribute than the second image. The whole network is jointly optimized via a unified end-to-end deep learning scheme. Extensive experiments on several benchmark datasets demonstrate the effectiveness of our approach against the state-of-the-art methods.} }

%0 Conference Paper %T Relative Attribute Learning with Deep Attentive Cross-image Representation %A Zeshang Zhang %A Yingming Li %A Zhongfei Zhang %B Proceedings of The 10th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2018 %E Jun Zhu %E Ichiro Takeuchi %F pmlr-v95-zhang18d %I PMLR %J Proceedings of Machine Learning Research %P 879--892 %U http://proceedings.mlr.press %V 95 %W PMLR %X In this paper, we study the relative attribute learning problem, which refers to comparing the strengths of a specific attribute between image pairs, with a new perspective of cross-image representation learning. In particular, we introduce a deep attentive cross-image representation learning (DACRL) model, which first extracts single-image representation with one shared subnetwork, and then learns attentive cross-image representation through considering the channel-wise attention of concatenated single-image feature maps. Taking a pair of images as input, DACRL outputs a posterior probability indicating whether the first image in the pair has a stronger presence of attribute than the second image. The whole network is jointly optimized via a unified end-to-end deep learning scheme. Extensive experiments on several benchmark datasets demonstrate the effectiveness of our approach against the state-of-the-art methods.