Abstract:
Person search has recently gained attention as the new task of jointly optimizing two tasks of person detection and person re-identification. Given the query person image...Show MoreMetadata
Abstract:
Person search has recently gained attention as the new task of jointly optimizing two tasks of person detection and person re-identification. Given the query person image to be searched, it can directly locate the position of the target person from panoramic surveillance images. Person search is more difficult but more practical and meaningful. In this paper, we propose a convolutional neural network for person search based on attention mechanism. Our network consists of two parts, the person proposal network and the identification network. Given a complete images gallery, the person proposal network generates persons’ bounding boxes, which are put into the identification network to extract features that are compared with the query person. Experiments on a large-scale benchmark dataset CUHK-SYSU demonstrate that our network gets better mean Average Precision result of 78.9%.
Published in: 2019 19th International Symposium on Communications and Information Technologies (ISCIT)
Date of Conference: 25-27 September 2019
Date Added to IEEE Xplore: 21 November 2019
ISBN Information: