Call for Participation
Recently, multi-camera visual surveillance has attracted much attention and been widely used in applications such as continuously tracking objects of interest, large-scale surveillance video retrieval, etc. The MCT challenge focuses on the problem of object tracking across multiple non-overlapping cameras. As shown in Figure 1, in the fields of view (FoVs) of all the cameras, every pedestrian must have a unique label or identity, no matter he enters the FoVs for the first time or re-appears for several times. Many researchers and organizations have done lots of work on this problem; however, it is very difficult to make comparisons between algorithms. Because there is few public benchmark datasets, most of the algorithms have been tested on different datasets. In the MCT challenge, all the participants must submit their tracking results on the same NLPR_MCT dataset using the same evaluation criterion.
Participants are encouraged to submit papers describing their submission methods to the workshop. The winner of the MCT challenge will give an oral presentation at the workshop, will receive a prize and will be invited to publish his method on a special issue of a top-rank journal.
- Deadline for Challenge Submission: EXTENDED to Sunday, 10th August 2014
- Notification of Results: Friday, 15th August 2014
- Date of Workshop: Friday, 12th September 2014
- The state of the art methods on NLPR_MCT dataset are reported now. 09.06.2017
- The evaluation kit is unpdated to version 3.0. 09.06.2017
- An alternative download of the NLPR_MCT dataset via Dropbox is available now. 03.06.2014
Prof. Kaiqi Huang, Email: firstname.lastname@example.org
Dr. Xiaotang Chen, Email: email@example.com
Weihua Chen, Email: firstname.lastname@example.org
Center for Research on Intelligent Perception and Computing (CRIPAC)
National Laboratory of Pattern Recognition(NLPR)
Institute of Automation, Chinese Academy of Sciences(CASIA)
Beijing 100190, China
Tel: +86 10 82544611, Fax: +86 10 82544485.