SemanticGenesis icon indicating copy to clipboard operation
SemanticGenesis copied to clipboard

How to train your model if I have already had the semantic features?

Open JiayinZheng opened this issue 3 years ago • 3 comments

How should I set the data_dir parameter if I already have the semantic_genesis_chest.ct in my directory? Thank you and looking forward to your reply. image

JiayinZheng avatar Jan 07 '22 10:01 JiayinZheng

Or how can I validate your results in LUNA dataset?

JiayinZheng avatar Jan 07 '22 14:01 JiayinZheng

How should I set the data_dir parameter if I already have the semantic_genesis_chest.ct in my directory? Thank you and looking forward to your reply.

You need to set the data_dir when you want to pre-train Semantic Genesis on your own dataset. In this case, data_dir should be set to the path of pre-training data.

fhaghighi avatar Jan 09 '22 05:01 fhaghighi

Or how can I validate your results in LUNA dataset?

If you want to directly evaluate the pre-trained Semantic Genesis on LUNA (or any other dataset), you first need to load the pre-trained model and then fine-tune it on the target task. The instructions for fine-tuning Semantic Genesis on any target task can be found under the “Fine-tune Semantic Genesis on your own target task” section in Pytorch and Keras directories.

fhaghighi avatar Jan 09 '22 05:01 fhaghighi