Pose recognition of giant pandas based on gradient shapes

In this paper, we proposed a novel approach towards the pose recognition of giant pandas based on the gradient shapes. The proposed approach is implemented in three main steps. First, the gradient contour is extracted by filtering the internal pixels in the gradient image. Second, a novel region partition method is used to measure the body weight of giant panda in different image directions. Third, the Fuzzy C-Means clustering is utilized to recognize various poses of giant panda based on the gradient shapes. Extensive experiments are carried out to validate the performance of the proposed method. The experimental results demonstrate that our method is quite promising.