avasalya/rpl repository overview

⁠This is a docker image for Automated Data Annotation for 6-DoF Object Pose Estimation

⁠Note: I am not the author of this paper or this work, my contribution solely lies on the development of this docker container.

original github: https://github.com/rohanpsingh/RapidPoseLabels⁠

original paper: https://arxiv.org/pdf/2011.03790.pdf⁠

⁠how to use

⁠tested on: Ubuntu 18.04, RTX2080Ti, Cuda 10.1, but also possible to use on other devices

⁠DIY: install Elastic Fusion https://github.com/avasalya/ElasticFusion -- since building ElasticFusion relies on specifying graphics card

1. pull docker image docker pull avasalya/rpl:cuda

make sure it is there, run docker images

1. run docker image as container sudo nvidia-docker run --gpus 0 -it -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=$DISPLAY -e NVIDIA_DRIVER_CAPABILITIES=all -h $HOSTNAME -v $HOME/.Xauthority:/home/USER/.Xauthority avasalya/rpl:cuda

2. once inside, you will have to copy your rosbag files from host PC to container, you can do that by using sudo docker cp /path_to_host_file **containerid**:rogbags/

3. follow further instructions from https://github.com/rohanpsingh/RapidPoseLabels⁠