Adenocarcinoma classification using deep neural networks
root: |data -> GlaS |trainingg_models |all codes should be here
#-------------------- Data set and trained models ------------------ GlaS Dataset can be downloaded from: https://warwick.ac.uk/fac/sci/dcs/research/tia/glascontest
Trained models can are provided in this link:
https://drive.google.com/open?id=1g6dhJ80zjqVA3hyQotxnbi2ZpLIzdlSl
#------------------------- Running instruction ---------------------
Please type this in bash for training and evaluation: bash run_exp.sh
Then you may run this for evaluation on test set: bash run_eval.sh
if you can not use bash you may use this command in command line for trainig a model: python modified_UNet_test.py experiment_num loss_type learning_rate
if you can not use bash you may use this command in command line, for evaluating the trained model: python evaluation.py experiment_num loss_type
Because we have uploaded our trained models with the experiment number of 1, in the run_exp.sh experiment_num is equal to 2 to not overwrite those files. If you wish to evaluate another experiment naturally you have to change experiment number in the run_eval.sh
#----------------------- Results and models -----------------------
all the models will be save in trained_models folder in a folder with a name corresponding to their loss type and experiment number.
all results will be saved in the same directory in the final_results folder.