https://github.com/wkentaro/label-fusion
Volumetric Fusion of Multiple Semantic Labels and Masks
https://github.com/wkentaro/label-fusion
3d-segmentation 3d-vision computer-vision semantic-segmentation
Last synced: 6 months ago
JSON representation
Volumetric Fusion of Multiple Semantic Labels and Masks
- Host: GitHub
- URL: https://github.com/wkentaro/label-fusion
- Owner: wkentaro
- License: mit
- Created: 2017-06-16T22:34:38.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2017-08-15T12:14:09.000Z (about 8 years ago)
- Last Synced: 2025-03-28T21:12:30.255Z (7 months ago)
- Topics: 3d-segmentation, 3d-vision, computer-vision, semantic-segmentation
- Language: C++
- Homepage:
- Size: 16.3 MB
- Stars: 19
- Watchers: 3
- Forks: 7
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# label-fusion: Volumetric Fusion of Multiple Semantic Labels
C++ code to fuse multiple object labels or mask into OctoMap,
which can be then used for 3d reconstruction of objects.
It works with and without depth inputs, so can be applied for depth insensible objects:
texture-less (for stereo), black (for ir), and transparent.
## Requirements
- [OpenCV](http://opencv.org) (tested with OpenCV 2.4.8)
- [Eigen](http://eigen.tuxfamily.org)
- [octomap (modified version)](https://github.com/wkentaro/octomap/tree/label_fusion) (automatically resolved by `build.sh`)
- [PCL](http://pointclouds.org) (tested with PCL 1.7.1)
## Installation
```bash
git clone https://github.com/wkentaro/label-fusion.git
cd label-fusion
./build.sh
```
## Demo
### Fusion of multiple labels
```bash
./demo.py label_view # see inputs
./demo.py label_fusion
./demo.py label_fusion --depth # use depth
```
### Fusion of multiple masks
We also support fusing multiple masks:
```bash
./demo.py mask_view # see inputs
./demo.py mask_fusion
./demo.py mask_fusion --depth # use depth
```
## License
MIT License (see `LICENSE` file).