An implementation of Kinect fusion (reconstruct surfaces from rgb and depth images) in pytorch
1.docker pull homagni/kinfusion:latest
2.git clone this repo and cd inside
(check the comments in scripts/start_container.bash and modify accordingly depending on where you want to attach the code, default is fine too)
3.cd scripts
chmod +x start_container.bash
./start_container.bash
(start visualizing a demo on frieberg dataset)
4.cd /slam #(the default location inside the container where you attached this github repo)
5.conda activate kinfu
6.python kinfu_gui.py --config configs/fr1_desk.yaml