update default params

Author: taigw

segmentation/JSRT/config/train_test.cfg

@@ -10,7 +10,6 @@ test_csv  = config/jsrt_test.csv
 
 train_batch_size = 4
 
-load_pixelwise_weight = False
 # modality number
 modal_num = 1
 
@@ -20,24 +19,13 @@ valid_transform = [NormalizeWithMeanStd, LabelConvert, LabelToProbability]
 test_transform  = [NormalizeWithMeanStd]
 
 NormalizeWithMeanStd_channels = [0]
-NormalizeWithMeanStd_mean = None
-NormalizeWithMeanStd_std  = None
-NormalizeWithMeanStd_mask = False
-NormalizeWithMeanStd_random_fill = False
-NormalizeWithMeanStd_inverse     = False
 
 LabelConvert_source_list = [0, 255]
 LabelConvert_target_list = [0, 1]
-LabelConvert_inverse = False
 
 RandomCrop_output_size = [240, 240]
-RandomCrop_foreground_focus = False
-RandomCrop_foreground_ratio = None
-RandomCrop_mask_label       = None
-RandomCrop_inverse     = False
 
 LabelToProbability_class_num = 2
-LabelToProbability_inverse   = False
 
 [network]
 # this section gives parameters for network
@@ -59,8 +47,6 @@ bilinear      = False
 gpus = [0]
 
 loss_type     = DiceLoss
-DiceLoss_enable_pixel_weight = False
-DiceLoss_enable_class_weight = False
 
 # for optimizers
 optimizer     = Adam
@@ -88,12 +74,6 @@ gpus       = [0]
 # checkpoint mode can be [0-latest, 1-best, 2-specified]
 ckpt_mode         = 0
 output_dir        = result
-evaluation_mode   = True
-test_time_dropout = False
-
-# use test time augmentation
-tta_mode              = 0
-infer_sliding_window  = False
 
 # convert the label of prediction output
 label_source = [0, 1]

segmentation/fetal_hc/README.md

@@ -26,7 +26,7 @@ pymic_net_run train config/train_test.cfg
 ![avg_loss](./picture/train_avg_loss.png)
 
 ## Testing and evaluation
-1. Run the following command to obtain segmentation results of testing images based on the best-performing checkpoint on the validation set. By default we use test time augmentation and sliding window inference to get better results. You can also edit the `testing` section of `config/train_test.cfg` to use other inference strategies.
+1. Run the following command to obtain segmentation results of testing images based on the best-performing checkpoint on the validation set. By default we use sliding window inference to get better results. You can also edit the `testing` section of `config/train_test.cfg` to use other inference strategies.
 
 ```bash
 mkdir result
@@ -39,6 +39,6 @@ pymic_net_run test config/train_test.cfg
 pymic_evaluate_seg config/evaluation.cfg
 ```
 
-The obtained average Dice score by default setting should be close to 97.12%. You can set `metric = assd` in `config/evaluation.cfg` and run the evaluation command again to get Average Symmetric Surface Distance (ASSD) evaluation results. 
+The obtained average Dice score by default setting should be close to 97.10%. You can set `metric = assd` in `config/evaluation.cfg` and run the evaluation command again to get Average Symmetric Surface Distance (ASSD) evaluation results. 
 
-3. Set `tta_mode = 0` and `infer_sliding_window  = False` in `config/train_test.cfg`, and run the testing and evaluation code again, we find that the average Dice will be decreased to around 96.86%.
+3. Set `tta_mode = 1` in `config/train_test.cfg` to enable test time augmentation, and run the testing and evaluation code again, we find that the average Dice will be increased to around 97.22%.

segmentation/fetal_hc/config/train_test.cfg

@@ -10,42 +10,27 @@ test_csv  = config/fetal_hc_test.csv
 
 train_batch_size = 4
 
-load_pixelwise_weight = False
 # modality number
 modal_num = 1
 
 # data transforms
 train_transform = [Rescale, RandomCrop, RandomFlip, NormalizeWithMeanStd, LabelConvert, LabelToProbability]
-valid_transform  = [Rescale, NormalizeWithMeanStd, LabelToProbability]
+valid_transform  = [Rescale, NormalizeWithMeanStd, LabelConvert, LabelToProbability]
 test_transform  = [Rescale, NormalizeWithMeanStd]
 
-Rescale_output_size = [256, 384]
-Rescale_inverse     = True
-
+Rescale_output_size    = [256, 384]
 RandomCrop_output_size = [224, 320]
-RandomCrop_foreground_focus = False
-RandomCrop_foreground_ratio = None
-RandomCrop_mask_label       = None
-RandomCrop_inverse     = False
 
 RandomFlip_flip_depth  = False
 RandomFlip_flip_height = True
 RandomFlip_flip_width  = True
-RandomFlip_inverse     = False 
 
 NormalizeWithMeanStd_channels = [0]
-NormalizeWithMeanStd_mean = None
-NormalizeWithMeanStd_std  = None
-NormalizeWithMeanStd_mask = False
-NormalizeWithMeanStd_random_fill = False
-NormalizeWithMeanStd_inverse     = False
 
 LabelConvert_source_list = [0, 255]
 LabelConvert_target_list = [0, 1]
-LabelConvert_inverse = False
 
 LabelToProbability_class_num = 2
-LabelToProbability_inverse   = False
 
 [network]
 # this section gives parameters for network
@@ -67,8 +52,6 @@ bilinear      = True
 gpus = [0]
 
 loss_type     = DiceLoss
-DiceLoss_enable_pixel_weight = False
-DiceLoss_enable_class_weight = False
 
 # for optimizers
 optimizer     = Adam
@@ -94,14 +77,13 @@ iter_save  = 5000
 gpus       = [0]
 
 # checkpoint mode can be [0-latest, 1-best, 2-specified]
-ckpt_mode         = 1
-output_dir        = result
-evaluation_mode   = True
-test_time_dropout = False
+ckpt_mode  = 1
+output_dir = result
 
 # use test time augmentation
-tta_mode              = 1
-infer_sliding_window  = True
+tta_mode = 0
+
+sliding_window_enable = True
 sliding_window_size   = [224, 320]
 sliding_window_stride = [224, 320]