This is the simulation code used for my thesis, Dataset and Evaluation of Self-Supervised Learning for Panoramic Depth Estimation.

Useful scripts are simulate.py, make_luts.py, stitch.py, and upload.py, all of which have argparse help except upload.py.

upload.py usage:

Takes either a list of recording directories, or "a/all <base_dir>" ("a" or "all") to upload all. If --remove is present, removed data files after upload. If --reupload is present, uploads files even if they have already been uploaded (upload file is present).

Carla constants are in carla_constants.py.

export.py exports depthmaps as mesh (.ply) files. The recording is hard coded in the script. make_image_sample.py, make_samples.py, and make_video_samples.py are utility scripts that aren't intended for anyone else's use, but the source may be informative. simulate_all.py is similar: it is used to run batches of simulations using hardcoded parameters.