As the important town of research on kinship recognition,
Northeastern University in US organized a face-based kinship recognition
competition in May 2017. This competition lasted three months and attracted
many teams from China, the United States, France, Israel, Algeria and other
countries and regions. Organizers held a workshop on "Recognize families
in the Wild 2017" at ACM multimedia, MM2017, where the winners were
announced and presented at a followed ceremony.
VIPL team won the first prize in the
competition. Li Yong, as representative of VIPL team, gave a 15-minute oral
report on the core algorithms of “Kinnet: fine to coarse deep metric learning
for kinship verification”at the workshop. The report received questions and
concerns from many peers, e.g., Lin Xi as a Hewlett-Packard engineer and Christopher
Mile as an engineer from US Homeland Security Department.
The competition results show that VIPL team ranked first (including in all seven subcompetitions), with an average accuracy rate of six percentage points higher than that of the second place (Figure 1). Joseph Robinson, as the chairman of RFIW 2017 workshop, rated it as "better than second place with a big margin." During award ceremony, Li Yong received the award on behalf of the VIPL team (Figure 3) and took a photo with the chairman Joseph Robinson (Figure 2).
Download: