Zero-Shot Object Detection with Textual Descriptions

• DOI: 10.1609/aaai.v33i01.33018690
• Publication Date: 2019-08-20T07:43:13Z
• Authors: Zhihui Li; Lina Yao; Xiaoqin Zhang; Xianzhi Wang; Salil Kanhere; Huaxiang Zhang
• Keywords: Benchmark (surveying); Zero (linguistics)

Object detection is important in real-world applications. Existing methods mainly focus on object with sufficient labelled training data or zero-shot only concept names. In this paper, we address the challenging problem of natural language description, which aims to simultaneously detect and recognize novel instances textual descriptions. We propose a deep learning framework jointly learn visual units, visual-unit attention word-level attention, are combined achieve word-proposal affinity by an element-wise multiplication. To best our knowledge, first work Since there no directly related literature, investigate plausible solutions based existing for fair comparison. conduct extensive experiments three benchmark datasets. The experimental results confirm superiority proposed model.