Learning Structured Semantic Embeddings for Visual Recognition (1706.01237v1)

Abstract: Numerous embedding models have been recently explored to incorporate semantic knowledge into visual recognition. Existing methods typically focus on minimizing the distance between the corresponding images and texts in the embedding space but do not explicitly optimize the underlying structure. Our key observation is that modeling the pairwise image-image relationship improves the discrimination ability of the embedding model. In this paper, we propose the structured discriminative and difference constraints to learn visual-semantic embeddings. First, we exploit the discriminative constraints to capture the intra- and inter-class relationships of image embeddings. The discriminative constraints encourage separability for image instances of different classes. Second, we align the difference vector between a pair of image embeddings with that of the corresponding word embeddings. The difference constraints help regularize image embeddings to preserve the semantic relationships among word embeddings. Extensive evaluations demonstrate the effectiveness of the proposed structured embeddings for single-label classification, multi-label classification, and zero-shot recognition.