Skip to content

README.md

Latest commit

 

History

History
27 lines (19 loc) · 847 Bytes

README.md

File metadata and controls

27 lines (19 loc) · 847 Bytes

Scan-rescan dataset

Preprocessing

Read the preprocessing readme on how to preprocess the data.

Training

The following command can be used to train/fine-tune the model on the dataset (please adjust the model.ckpt_path and data.dir accordingly):

# unet from scratch
rescan_seg_train data.dir=~/.cache/cinema_datasets/rescan/processed model.name=unet
# from scratch
rescan_seg_train data.dir=~/.cache/cinema_datasets/rescan/processed model.name=convunetr
# fine-tune
rescan_seg_train data.dir=~/.cache/cinema_datasets/rescan/processed model.name=convunetr model.ckpt_path=

Evaluation

The following command can be used to evaluate the model (please adjust the model.ckpt_path accordingly):

rescan_seg_eval --data_dir ~/.cache/cinema_datasets/rescan/processed --ckpt_path