School of Internet of Things, Jiangnan University
The National Natural Science Foundation of China under Grant NO. 61573167、Fund for “Integration of Cloud Computering and Big Data” of Innovation of Science and Education under Grant NO. 2017A13055
In most target tracking algorithms based on siamese network, the feature extraction ability of the backbone network is weak and the template can’t adapt to the change of target. A deformable convolution siamese network algorithm(DCSiam) is proposed based on SiamFC algorithm. Firstly, to enhance the capability of deep semantic feature extraction of backbone network, the deformable convolution module is used to learn the adaptive offset of multi-layer feature data in different directions, and the effective receptive field in the convolution process is increased. The final response map is obtained by multi-layer deformable cross-correlation fusion. Finally, an online template updating strategy with high confidence is adopted. The peak sidelobe ratio an the maximum value of the response graph were calculated every fixed frame as the basis for updating, and the features were fused were in a weighted way to update the template. The tracking performance of the proposed algorithm is evaluated using four common benchmark datasets:OTB2013、OTB2015、VOT2016 and VOT2017. The experimental results show that the overall accuracy and success rate of the DCSiam algorithm are increased by 9.5% and 7.5% respectively compared with the baseline on the OTB2015 data set, which well realizes the target tracking in complex situations and verifies and verifies the effectiveness of the proposed algorithm.