MHSAN: Multi-Head Self-Attention Network for Visual Semantic Embedding.

Geondo Park, Chihye Han, Daeshik Kim, Wonjun Yoon

Browse the full WACV paper archive.