DS-SLAM

## Overview - [Official Github](https://github.com/ivipsourcecode/DS-SLAM.git) - [My forked Source Code](https://github.com/yubaoliu/DS-SLAM.git) ## Paper Yu, C., Liu, Z., Liu, X. J., Xie, F., Yang, Y., Wei, Q., & Fei, Q. (2018). DS-SLAM: A Semantic Visual SLAM towards Dynamic Environments. IEEE International Conference on Intelligent Robots and Systems, 1168–1174. https://doi.org/10.1109/IROS.2018.8593691 ## Demo ## Build ### Environment ```sh Ubuntu 16.04 ROS Kinetic Cuda 10.0 Cudnn 7 Caffe (SegNet) Eigen 3 OpenCV 3 [optional] docker: 19.03.12 ``` ### Add to ROS PACKAGEPATH ```sh export ROS_PACKAGE_PATH=$ROS_PACKAGE_PATH:$HOME/catkin_ws/src/DS-SLAM/Examples/ROS/ORB_SLAM2_PointMap_SegNetM ``` ### Build and install caffe-segnet-cudnn5 ```sh git clone https://github.com/yubaoliu/caffe-segnet.git cd SegNet/caffe-segnet mkdir build cd build cmake .. -CMAKE_INSTALL_PREFIX=/usr/local make & make install ``` ### Build DS-SLAM ```sh bash DS_SLAM_BUILD.sh ``` ## Obtain Caffe model ```sh http://mi.eng.cam.ac.uk/~agk34/resources/SegNet/segnet_pascal.caffemodel ``` ## Run Demo Please customize the directory path of dataset and caffe model before run the launch file. ```sh roslaunch DS_SLAM_TUM3.launch ```