The team of professor Ruimin Hu won the only best paper award of PCM 2014.

The team of professor Ruimin Hu won the only best paper award of PCM 2014. The 15th general assembly was held in Malaysia la yue state capital, Kuching on December 1-4, 2014. After the fierce conflicts in paper review and reports, the paper “Region-based Interactive Ranking Optimization for Person Re-identification” submitted by PhD student Zheng Wang, professor Ruimin Hu, and Dr Liang Chao from computer college of wuhan university won the only best paper award of PCM. Review committees all agreed that “The method is very simple and fast, and can be applied in the real world to solve the important problem effectively.”。

Pedestrian re-identification refers to retrieving the same objects among the vast amount of data recorded by different terminal equipments. Due to the difference in video recording equipment, shooting angle and light conditions, the retrieval quality would decrease seriously. In recent years, multimedia intelligent retrieval technology is developing rapidly, as a key technology in the field of intelligent retrieval, the pedestrian re-identification technology has aroused wide attention in city security and criminal investigation, consumer behavior perception and analysis, which is also key technology of big data analysis recommended by big data analysis report in the president of the United States advisory committee of science and technology in 2014. In the field of city security, video monitoring technologies are playing an increasingly important role in the prevention and control of crime and illegal activities. Existing suspected target screening work is mainly rely on artificial process, which requires many manpower material resources, and simultaneously, artificial screening approach under the environment of mass surveillance video is inefficient. How to utilize computers to complete automatic analysis and comparison quickly, how to effectively improve the search speed and accuracy of suspected target, have become a hot research topic in the field of computer vision and multimedia retrieval.

Under the continued support of national natural science foundation, 973 subprojects, ministry of public security key scientific research projects, the team of professor Ruimin Hu paid their attention to surveillance video retrieval of specific pedestrian target in multi-camera environment, discovered that both the positive samples and negative samples were sparse, proposed the area-based interactive optimization method, and realized the query re-optimization of pedestrian re-identification. The CMC(Cumulative Matching Characteristic) on benchmark datasets has a promotion of 20%, while the retrieval time reduced to half its former.