報告題目:Multiple mice tracking using deep learning.
報告人: Dr. Huiyu (Joe) Zhou,Reader, Department of Informatics, University of Leicester, UK.
摘要:The study of mouse social behaviours has beenincreasingly undertaken in neuroscience research. However, automatedquantification of mouse behaviours from the videos of interacting mice is stilla challenging problem, where object tracking plays a key role in locating micein their living spaces. In this talk, we propose a novel method to continuouslytrack several mice and individual parts without requiring any specific tagging.Firstly, we propose an efficient and robust deep learning based mouse partdetection scheme to generate part candidates. Subsequently, we propose a novelBayesian Integer Linear Programming Model that jointly assigns the partcandidates to individual targets with necessary geometric constraints whilstestablishing pair-wise association between the detected parts. There is nopublicly available dataset in the research community that provides aquantitative test-bed for the part detection and tracking of multiple mice, andwe here introduce a new challenging Multi-Mice PartsTrack dataset that is madeof complex behaviours and actions. Finally, we evaluate our proposed approachagainst several baselines on our new datasets, where the results show that ourmethod outperforms the other state-of-the-art approaches in terms of accuracy.
Dr. Huiyu Zhou received a Bachelor of Engineering degree in RadioTechnology from Huazhong University of Science and Technology of China, and aMaster of Science degree in Biomedical Engineering from University of Dundee ofUnited Kingdom, respectively. He was awarded a Doctor of Philosophy degree inComputer Vision from Heriot-Watt University, Edinburgh, United Kingdom. Dr. Zhou currently is a Reader at Department ofInformatics, University of Leicester, United Kingdom. He haspublished over 200 peer-reviewed papers in the field. He was the recipient of"CVIU 2012 Most Cited Paper Award", “ICPRAM 2016 Best PaperAward” and was nominated for “ICPRAM 2017 Best Student Paper Award” and "MBEC2006 Nightingale Prize". Four of his papers recently published byElsevier were ranked as the ScienceDirect Top 25 Articles. Dr. Zhouserves as the Editor-in-Chief of Recent Advances in Electrical & ElectronicEngineering and Associate Editor of "IEEE Transaction on Human-MachineSystems", and is on the Editorial Boards of several refereed journals. He is one of theTechnical Committee of “Information Assurance & IntelligentMultimedia-Mobile Communication in IEEE SMC Society”, “Robotics Task Force” and“Biometrics Task Force” of the Intelligent Systems Applications TechnicalCommittee, IEEE Computational Intelligence Society. He has given over 50 invited talks at internationalconferences, industry and universities, and has served as a chair for 30international conferences and workshops. His research work has been or is beingsupported by UK EPSRC, MRC, EU, Royal Society, Leverhulme Trust, Puffin Trust,Alzheimer’s Research UK, Invest NI and industry.