## Data preparation
Data is prepared using the prepare_s3dis_label.py.
For training on area 2:
python s3dis_seg.py --rootdir path_to_data_processed/ --area 2 --savedir path_to_save_directory
To train without color information (used for fusion):
python s3dis_seg.py --rootdir path_to_data_processed/ --area 2 --savedir path_to_save_directory --nocolor
For testing on area 2:
python s3dis_seg.py --rootdir path_to_data_processed --area 2 --savedir path_to_save_directory --test
If the --nocolor option was used at training, it should be also used during test:
python s3dis_seg.py --rootdir path_to_data_processed --area 2 --savedir path_to_save_directory --nocolor --test
After training of both color model and segmentation model, you can train a fusion model with:
python s3dis_seg_fusion.py --rootdir path_to_data_processed --area 2 --savedir path_to_save_directory --model_rgb path_to_model_rgb_dir --model_noc path_to_model_nocolor_dir
python s3dis_seg_fusion.py --rootdir path_to_data_processed --area 2 --savedir path_to_fusion_model_dir --model_rgb path_to_model_rgb_dir --model_noc path_to_model_nocolor_dir --test
python s3dis_eval.py --datafolder path_to_data_processed --predfolder pathèto_model --area 2