VIPL researchers participated in the ICCV2015 computer Vision Top Conference Competition "ICCV 2015 ChaLearn looking at people Challenge" (http: / gesture.chalearn.org /), held in Chile On December 12, 2015. The competition is divided into two subcontests: the Track 1 of Apparent Age estimation contest and the Track 2 of Culture event recognition contest. The Track-1 Apparent Age estimation contest is the first international competition on the topic of social age estimation, which is mainly aimed at the problem of social age estimation in real scenes. The Track-2 Culture event recognition contest sets the task of classifying 100 cultural events from a static image.
The organizers finally announced the results of the top 10 entries in the two subcontests. In Track-2 Culture event recognition contest, VIPL PHD students of Liu Mengyi, Liu Xin and Li Yan designed the deep convolution network based on hierarchical feature extraction combined with classical subspace learning method, and finally won the championship. In Track-1 parent Age estimation contest, VIPL PHD students of Liu Xin, Li Shaoxin, Zhang Jie and Wu Shuzhe designed end-to-end learning methods of age classifier and regressor based on super large scale convolution neural network under the guidance of task group teachers. In the end, they won runner-up. (For details, see http: / gesture.chalearn.org /)
Download: