基于改进深度孪生网络的分类器及其应用-计算机工程与应用.PDFVIP

Computer Engineering and Applications 计算机工程与应用 2018 ，54（10） 19 基于改进深度孪生网络的分类器及其应用 用 1 1，2 1，2 沈 雁 ，王 环 ，戴瑜兴 应 1 1，2 1，2 SHEN Yan , WANG Huan , DAI Yuxing 与 1. 湖南大学 电气与信息工程学院，长沙 410082 程 2.温州大学 数理与电子信息工程学院，浙江 温州 325035 g r 1.College of Electrical and Information Engineering, Hunan University, Changsha 410082, China 工 o 2.College of Mathematics, Physics and Electronic Information Engineering, Wenzhou University, Wenzhou, Zhejiang . 325035, China 机 j a 算 e SHEN Yan, WANG Huan, DAI Yuxing. Deep siamese network-based classifier and its application. Computer Engi- c . neering and Applications, 2018, 54 （10）：19-25. 计 w Abstract ：Siamese neural network consists of twin networks which share the same parameters, and can map nonlinear w data with high dimension onto a low dimension and easy separable feature space. By taking use of its excellent perfor- w mance on similarity computing, an efficient classifier based on Siamese network is proposed. Convolutional Neural Net- works （CNNs ）are used as building blocks, combined with max-pooling and dropout techniques, to construct a multi-scale CNNs for feature extraction. A spatial transformer network is used as a supplementary to promote the accuracy of the clas- sifier. Through experiments on traffic sign data GTSRB, the classifier obtains 99.40% accuracy rate for this recognition benchmark. The proposed method for classification has virtues of simple structure, high accuracy, short training and fast recognition.