Paper ID: 2111.12757

ACNet: Approaching-and-Centralizing Network for Zero-Shot Sketch-Based Image Retrieval

Hao Ren, Ziqiang Zheng, Yang Wu, Hong Lu, Yang Yang, Ying Shan, Sai-Kit Yeung

The huge domain gap between sketches and photos and the highly abstract sketch representations pose challenges for sketch-based image retrieval (\underline{SBIR}). The zero-shot sketch-based image retrieval (\underline{ZS-SBIR}) is more generic and practical but poses an even greater challenge because of the additional knowledge gap between the seen and unseen categories. To simultaneously mitigate both gaps, we propose an \textbf{A}pproaching-and-\textbf{C}entralizing \textbf{Net}work (termed "\textbf{ACNet}") to jointly optimize sketch-to-photo synthesis and the image retrieval. The retrieval module guides the synthesis module to generate large amounts of diverse photo-like images which gradually approach the photo domain, and thus better serve the retrieval module than ever to learn domain-agnostic representations and category-agnostic common knowledge for generalizing to unseen categories. These diverse images generated with retrieval guidance can effectively alleviate the overfitting problem troubling concrete category-specific training samples with high gradients. We also discover the use of proxy-based NormSoftmax loss is effective in the zero-shot setting because its centralizing effect can stabilize our joint training and promote the generalization ability to unseen categories. Our approach is simple yet effective, which achieves state-of-the-art performance on two widely used ZS-SBIR datasets and surpasses previous methods by a large margin.

Submitted: Nov 24, 2021