Re-Identification Method of Siberian Tiger Based on Adaptive Regularization

doi:10.3778/j.issn.1002-8331.2012-0143

Abstract

Abstract: As the number of Siberian tigers continues to decrease, it becomes very meaningful to identify a single tiger for protection and tracking. Therefore, a Siberian tiger re-identification network model based on local partial blocks and an adaptive L2 regularization method（part- based convolutional baseline-adaptiveL2, PCB-AL2） is adopted to solve the problem of difficulty in re-identification of Siberian tiger in natural environment. The adaptive L2 regularization factor is adaptively updated through backpropagation, which is achieved by using the regularization factor as a trainable variable. Aiming at the characteristics of tigers relying on body stripes to distinguish, a two-branch network structure is adopted：local branch and global branch. The network relies on local features to guide global feature learning. Experimental results show that comparing with PPbM-a, PPbM-b and PPGNet on the ATRW data set, it is concluded that mAP reaches 92.1% in a single-camera environment and 75.1% in a cross-camera environment.

Key words: re-identification, residual network, adaptive L2 regularization, characteristics of fusion

摘要： 随着东北虎数量不断减少，识别单只老虎进而做出保护和追踪变得很有意义，故采用了一种基于局部分块和自适应L2正则化方法的东北虎重识别网络模型（part-based convolutional baseline-adaptiveL2，PCB-AL2）以解决在自然环境下东北虎重识别困难等问题。自适应L2正则化因子通过反向传播进行自适应更新，这是通过将正则化因子作为可训练的变量来实现的。针对老虎依靠身体条纹分辨的特点，采用一种双分支网络结构：局部分支和全局分支，网络依靠局部特征指导全局特征学习。实验结果表明，在ATRW数据集上与PPbM-a、PPbM-b以及PPGNet对比得出结论，在单摄像头环境下mAP达到了92.1%，跨摄像头环境下mAP达到75.1%。

关键词: 重识别, 残差网络, 自适应L2正则化, 特征融合

YU Huiling, QIAN Chengshuai. Re-Identification Method of Siberian Tiger Based on Adaptive Regularization[J]. Computer Engineering and Applications, 2022, 58(8): 191-197.

于慧伶, 钱成帅. 基于自适应正则化的东北虎重识别方法[J]. 计算机工程与应用, 2022, 58(8): 191-197.

References

[1] ARZOUMANIAN Z，HOLMBERG J，NORMAN B，et al.An astronomical pattern-matching algorithm for computer-aided identification of whale sharks Rhincodon typus[J].Journal of Applied Ecology，2005，42（6）：999-1011.
[2] ARDOVINI A，CINQUE L，SANGINETO E，et al.Iden-tifying elephant photos by multi-curve matching[J].Pattern Recognition，2008，41（6）：1867-1877.
[3] CARTER S J，BELL I，MILLER J，et al.Automated marine turtle photograph identification using artificial neural networks，with application to green turtles[J].Journal of Experimental Marine Biology and Ecology，2014，452：105-110.
[4] WEIDEMAN H J，JABLONS Z M，HOLMBERG J，et al.Integral curvature representation and matching algorithms for identification of dolphins and whales[J].arXiv：1708.
07785，2017.
[5] KORSCHENS M，BARZ B，DENZLER J.Towards automatic identification of elephants in the wild[J].arXiv：1812.04418，2018.
[6] LI S，LI J，LIN W，et al.Amur tiger re-identification in the wild[J].arXiv：1906.05586，2019.
[7] LIU C，ZHANG R，GUO L，et al.Part-pose guided amur tiger re-identification[C]//International Conference on Computer Vision，2019.
[8] GUO J F，PANG Z Q，YU M，et al.A novel pedestrian reidentification method based on a multiview generative adversarial network[J].IEEE Access，2020，8：181943-181954.
[9] GUO J F，PANG Z Q，YANG F，et al.Study on the method of fundus image generation based on improved GAN[J].Mathematical Problems in Engineering，2020：6309596.
[10] GENG M Y，WANG Y W，XIONG T，et al.Deep transfer learning for person re-identification[J].arXiv：1611.
05244，2016.
[11] LIN Y T，ZHENG L，ZHENG Z D，et al.Improving person re-identification by attribute and identity learning[J].Pattern Recognition，2019，95：151-161.
[12] ZHENG L，YANG Y，HAUPTMANN A G.Person re-identification：past，present and future[J].arXiv：1610.02984，2016.
[13] ZHENG Z D，ZHENG L，YANG Y.Pedestrian alignment network for large-scale person re-identification[J].IEEE Transactions on Circuits and Systems for Video Technol-ogy，2019，29（10）：3037-3045.
[14] VARIOR R R，SHUAI B，LU J，et al.A Siamese long short-term memory architecture for human re-identification[J].arXiv：1607.08381，2016.
[15] ZHENG L，HUANG Y，LU H，et al.Pose-invariant embedding for deep person re-identification[J].IEEE Transactions on Image Processing，2019，28（9）：4500-4509.
[16] ZHAO H Y，TIAN M Q，SUN S Y，et al.Spindle net：person re-identification with human body region guided feature decomposition and fusion[C]//IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2017.
[17] WANG G S，YUAN Y F，CHEN X，et al.Learning discriminative features with multiple granularities for person re-identification[C]//International Conference on Multimedia，2018.
[18] VARIOR R R，HALOI M，WANG G.Gated Siamese convolutional neural network architecture for human re-identification[C]//European Conference on Computer Vision，2016：791-808.
[19] SCHROFF F，KALENICHENKO D，PHILBIN J.Facenet：a unified embedding for face recognition and clustering[C]//IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2015.
[20] LIU H，FENG J，QI M，et al.End-to-end comparative attention networks for person re-identification[J].IEEE Transactions on Image Processing，2017，26（7）：3492-3506.
[21] CHENG D，GONG Y，ZHOU S，et al.Person re-identification by multi-channel parts-based cnn with improved triplet loss function[C]//IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2016.
[22] CHEN W，CHEN X，ZHANG J，et al.Beyond triplet loss：a deep quadruplet network for person re-identification[C]//IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2017.
[23] CHEN D，XU D，LI H，et al.Group consistent similarity learning via deep CRF for person re-identification[C]//IEEE Conference on Computer Vision and Pattern Recognition（CVPR），2018.
[24] LAARHOVEN T V.L2 regularization versus batch and weight normalization[J].arXiv：1706.05350，2017.
[25] IOFFE S，SZEGEDY C.Batch normalization：accelerating deep network training by reducing internal covariate shift[J].arXiv：1502.03167，2015.
[26] SALIMANS T，KINGMA D P.Weight normalization：a simple reparameterization to accelerate training of deep neural networks[C]//Advances in Neural Information Processing Systems，2016：901-909.
[27] HOFFER E，BANNER R，GOLAN I，et al.Norm matters：efficient and accurate normalization schemes in deep networks[C]//Advances in Neural Information Processing Systems，2018：2160-2170.
[28] LOSHCHILOV I，HUTTER F.Decoupled weight decay regularization[J].arXiv：1711.05101，2017.
[29] LEWKOWYCZ A，GUR-ARI G.On the training dynamics of deep networks with L2 regularization[J].arXiv：2006.
08643，2020.
[30] SUN Y，ZHENG L，YANG Y，et al.Beyond part models：person retrieval with refined part pooling（and a strong convolutional baseline）[C]//European Conference on Computer Vision（ECCV），2018：480-496.
[31] KRIZHEVSKY A，HINTON G.Convolutional deep belief networks on cifar-10[Z].2010.
[32] ALEXANDER H，LUCAS B，BASTIAN L.In defense of the triplet loss for person reidentification[J].arXiv：1703.
07737，2017.
[33] NI X Y，FANG L，HUTTUNEN H.AdaptiveReID：adaptive L2 regularization in person re-identification[J].arXiv：2007.
07875，2020.
[34] ZHONG Z，ZHENG L，KANG G，et al.Random erasing data augmentation[C]//Proceedings of the AAAI Conference on Artificial Intelligence，2020，34（7）：13001-13008.
[35] ZHONG Z，ZHENG L，CAO D，et al.Re-ranking person re-identification with k-reciprocal encoding[C]//IEEE Conference on Computer Vision and Pattern Recognition，2017：1318-1327.