Two-stage zero-shot sparse hashing with missing labels for cross-modal retrieval

Kailing Yong, Zhenqiu Shu, Hongbin Wang, Zhengtao Yu

Abstract: Highlights•Each stage operates independently, resulting in a concise and efficient algorithm.•Our approach emphasizes mitigating the impact of missing labels.•We train a classifier to predict labels for instances with missing labels.•We jointly utilize cluster-wise similarity and sample-wise fine-grained similarity.•We employ high-dimensional sparse hash codes to improve the discriminative ability.