README.md

preprocess Replica

You can get the data I have processed in GoogleDrive

download the original Replica Dataset

Refer to the official repo

prepare

# download synsin to get the scene_episodes of replica.
git clone https://github.com/facebookresearch/synsin.git

remember to set the following directory path to your own in class Options of main_wide.py:

self.scenes_dir: Replica data storage directory

self.basedir: the base directory where you put synsin and mp3d.yaml

self.episodes_dir: the specific directory where you put scene episodes.

final_save_dir: the directory to save processed data

Replica data structure:

${self.scenes_dir}
     | -- apartment_0
             | -- glass.sur
             | -- mesh.ply
             ...
     | -- apartment_1
             | -- ...
             | -- ...

preprocess

run script to preprocess the original data to get 360-degree replica data. bash main_wide.sh

Acknowledgements

The code is supported by the author of SOMSI and OmniSyn.