The academic paper "Attend and Align: Improving Deep Representations with Feature Alignment Layer for Person Retrieval" written by postgraduate Xu Qin in the team of Professor Wang Shengjin of the Research Center for Media Big-data Cognitive Computing of the Department of Electronic Engineering of our university won the 24th International Conference on Pattern Recognition (ICPR2018) Best Scientific Paper Award.
The International Conference on Pattern Recognition (ICPR) is one of the most influential academic conferences in the world in the international pattern recognition and machine learning field organized by the International Association for Pattern Recognition (IAPR), and is widely valued by experts and scholars in this research field worldwide. ICPR started from 1972, and is a flagship academic conference in the pattern recognition field organized by IAPR. It is held every two years, and the host country or region is decided by the council of the IAPR through secret ballot four years in advance.
The 24th International Conference on Pattern Recognition (ICPR 2018) was held from August 20 to August 24 at the China National Convention Center; Beijing, China, and this was the first time for ICPR to be held in the mainland China since its inception of over 40 years. The conference received a total of 1,258 contributions, and 125 oral papers were accepted, with an acceptance rate of 9.9%. This paper was accepted as oral and won the Best Scientific Paper Award. A total of 6 papers won the award at this conference. A total of over 500 representatives attended the meeting, including nearly 100 domestic representatives and over 400 overseas ones.
The team of Professor Wang Shengjin has international leading research results in the Person ReID research field, and the academic papers cover computer vision, AI top journals (TPAMI, TIP), and top international conferences (ICCV, CVPR, ECCV). MIT Technology Review spoke highly of the team. Meanwhile, the team has contributed 3 large-scale public image datasets: Market-1501, iLID-VID and video dataset MARS in the Person ReID research field. Among them, the downloads of iLID-VID have exceeded 3,700 times. The SVDNet method paper was selected as ICCV2017 spotlight, and the RPP method broke the highest level in the world. Presently, relevant technologies have been tested in the security field.
