SUNRGBD

Python tool to curate the SUNRGBD database for semantic segmentation

Download the SUNRGBD database as well as the toolbox to get the labels:

wget http://rgbd.cs.princeton.edu/data/SUNRGBD.zip
unzip SUNRGBD.zip
wget http://rgbd.cs.princeton.edu/data/SUNRGBDtoolbox.zip
unzip SUNRGBDtoolbox.zip

The parsing of the database is based on the SunRGBDDataset class which holds the number of labels as a class attribute (n_label=38), and the relationship between the label code (an integer between 0 and 37), and the label meaning ('lamp', 'bed', ...) in the instance attribute self.label_name.

A curated database can be created using:

python scripts/parse_database.py --path_2_database {path_2_database} --path_2_toolbox {path_2_toolbox} --output_dir {output_dir} --resize --nx 128 --ny 128 --grayscale

Which will create a database with grayscale images of size 128x128. It will save the image in an image folder in output_dir, and the labels in a label folder in output_dir. The database will be 170M.

Example images

References

Song, S., Lichtenberg, S.P. and Xiao, J., 2015. Sun rgb-d: A rgb-d scene understanding benchmark suite. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 567-576).