Towards Semantic Embedding in Visual Vocabulary

Visual vocabulary serves as a fundamental component in many computer vision tasks, such as object recognition, visual search, and scene modeling. While state-of-the-art approaches build visual voca