This is my graduation project.

======= #This semantic mapping code is my graduation project

Some results

Models

• Models files are not here, you can choose to modify this file:

//segnet.cpp
    std::string model_file = "../models/segnet_model_driving_webdemo.prototxt";
    std::string trained_file = "../models/segnet_weights_driving_webdemo.caffemodel";
    std::string label_file = "../models/semantic12.txt";

• The semantic segmentation methon is based on Alex Kendall's work

• Caffe Segnet is a modified version of Caffe which supports the SegNet architecture

• The mapping architecture is based on GaoXiang's work and ORB-SLAM.

---

Getting Started

If you would like to try out this code , you should satisfy these requirements first.

#OpenCV I choose to use OpenCV to manipulate images and features. Dowload and install instructions can be found at: http://opencv.org. Required at OpenCV 2.4.x.Note that my code is not compatible with OpenCV3.0

Eigen 3.0+

PCL 1.7

DBoW2(for loop closure) and g2o(for solving pnp)