Interactive product image search with complex scenes

Online shopping becomes a convenient way for billions for web users to purchase the products, especially clothes. A large portion of the product images are various types of apparel often including a human model from cluttered and non-uniform natural backgrounds, which makes visual product search a challenging task. In this work, we propose an approach for interactive product image search with complex scenes, which combines the interactive image segmentation for query images, and efficient graph-based principal object extraction for backend image database to extract the foreground objects, respectively. Experiments on a large scale dataset with 1.36 million product images crawled from Taobao demonstrate the effectiveness of the proposed solution.

[1]  Lifeng Sun,et al.  Auto-cut for web images , 2009, MM '09.

[2]  Chong-Wah Ngo,et al.  On the Annotation of Web Videos by Efficient Near-Duplicate Search , 2010, IEEE Transactions on Multimedia.

[3]  Daniel P. Huttenlocher,et al.  Efficient Graph-Based Image Segmentation , 2004, International Journal of Computer Vision.

[4]  Andrew Blake,et al.  "GrabCut" , 2004, ACM Trans. Graph..

[5]  Nenghai Yu,et al.  iLike: integrating visual and textual features for vertical search , 2010, ACM Multimedia.

[6]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[7]  Jitendra Malik,et al.  Normalized cuts and image segmentation , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[8]  Haojie Li,et al.  iSearch: towards precise retrieval of item image , 2011, ICIMCS '11.

[9]  Qiang Peng,et al.  Principal object detection towards product image search , 2012, 2012 International Conference on Audio, Language and Image Processing.