Monte-Carlo MapNet
Manohar Kuse mpkuse@connect.ust.hk
Trains a MapNet with Monte-Carlo process on a 3D model With intensity perfurbation, possible to have augmented features for training.
Remake of the original mc_train. More organized code and trying to run on the larger 3D model
Manohar Kuse, Sunil Prasad Jaiswal, Shaojie Shen, **Deep-mapnets: A Residual Network for 3D Environment Representation ** 2017 IEEE International Conference on Image Processing (ICIP 2017), Beijing, China, 17-20 September 2017
Prerequistite Packages
- Caffe
- Panda3d
- OpenCV
- Numpy
Rendering from 3D model
The 3D model depected below is available for download from here (4.2 GB). It is a model of HKUST, Hong Kong (GPS-lat:22.335162 GPS-long:114.263035)
If you use our model for training, do site us. The model was created with Altizure.
