Overfitting Reduction of Pose Estimation for Deep Learning Visual Odometry

作     者：Xiaohan Yang Xiaojuan Li Yong Guan Jiadong Song Rui Wang Xiaohan Yang;Xiaojuan Li;Yong Guan;Jiadong Song;Rui Wang

作者机构：Information Engineering CollegeCapital Normal UniversityBeijing 100048China Beijing Engineering Research Center of High Reliable Embedded System Beijing Advanced Innovation Center for Imaging Theory and Technology Machinery Industry Information CenterBeijing 100823China Beijing Key Laboratory of Light Industrial Robots and Safety Verification 

出 版 物：《China Communications》 (中国通信（英文版）)

年 卷 期：2020年第17卷第6期

页      面：196-210页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 081104[工学-模式识别与智能系统] 08[工学] 080203[工学-机械设计及理论] 0835[工学-软件工程] 0802[工学-机械工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：supported by National Key R&D Plan(2017YFB1301104),NSFC(61877040,61772351) Sci-Tech Innovation Fundamental Scientific Research Funds(025195305000)(19210010005),academy for multidisciplinary study of Capital Normal University 

主　　题：visual odometry neural network pose estimation bayesian distribution overfitting 

摘      要：Error or drift is frequently produced in pose estimation based on geometricfeature detection and trackingmonocular visual odometry(VO)when the speed of camera movement exceeds 1.5 m/***,in most VO methods based on deep learning,weight factors are in the form of fixed values,which are easy to lead to overfitting.A new measurement system,for monocular visual odometry,named Deep Learning Visual Odometry(DLVO),is proposed based on neural *** this system,Convolutional Neural Network(CNN)is used to extract feature and perform feature ***,Recurrent Neural Network(RNN)is used for sequence modeling to estimate camera’s 6-dof *** of fixed weight values of CNN,Bayesian distribution of weight factors are introduced in order to effectively solve the problem of network *** 18,726 frame images in KITTI dataset are used for training *** system can increase the generalization ability of network model in prediction *** with original Recurrent Convolutional Neural Network(RCNN),our method can reduce the loss of test model by 5.33%.And it’s an effective method in improving the robustness of translation and rotation information than traditional VO methods.