This is a C++ implementation of Fast Depth Densification for Occlusion-aware Augmented Reality (SIGGRAPH ASIA2018) and its python implementation.
The code generates dense depth map from sparse depth points of visual SLAM by using variational method.
Sample data is from AR-Depth.
- OpenCV (>3.2)
- Eigen 3.3.5
-
Download the repository.
git clone https://github.com/muskie82/AR-Depth-cpp.git
-
Build
cd AR-Depth-cpp mkdir build cd build cmake .. make -j4
You can run sample program
./AR_Depth /path/to/sample_data/frames /path/to/sample_data/reconsrtruction
- Fast Depth Densification for Occlusion-aware Augmented Reality, Aleksander Holynski and Johannes Kopf, IACM Transactions on Graphics (Proc. SIGGRAPH Asia). (https://homes.cs.washington.edu/~holynski/publications/occlusion/index.html)
- AR-Depth, (https://github.com/facebookresearch/AR-Depth)
GPLv3 license.