Skip to content

Annotate RGB-D scenes with background/foreground labels using ground truth 3D poses

Notifications You must be signed in to change notification settings

andersgb1/scene-annotator

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 

Repository files navigation

scene-annotator

Annotate an RGB-D scene with background/foreground labels using ground truth 3D poses.

Make sure to install CoViS from here and don't forget to compile the Python bindings using make covis_python.

You need the following Python dependencies for this script:

  • numpy
  • scikit-image

And finally you need to make sure that CoViS is built "one folder up" relative to this directory - otherwise, you have to change the line below the import statement near the top of scene_annotator.py to append the correct path to the bindings.

Testing

The script is tailored to load test sequences from the SIXD Challenge datasets. If you download one of the datasets from the webpage (you only need the models.zip and test.zip files from each dataset), you need to tell the script where to find the root of the dataset, what object to annotate, what scene sequence to process, and finally where to output the training data:

python scene_annotator.py /path/to/sixd_dataset/ --object=models/obj_01.ply --scene-dir=test/01 --output-dir=./output

About

Annotate RGB-D scenes with background/foreground labels using ground truth 3D poses

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages