clean up scripts

pixel_decoder/predict.py

@@ -17,16 +17,14 @@
 from pixel_decoder.utils import stats_data, open_image, preprocess_inputs_std, cache_stats
 from pixel_decoder.resnet_unet import get_resnet_unet
 
-def predict(origin_shape, imgs_folder, models_folder, pred_folder, origin_shape_no=256, border_no=32, channel_no=3, model_id='resnet_unet'):
+def predict(origin_shape, imgs_folder, models_folder, pred_folder, origin_shape_no, border_no, channel_no, model_id):
     origin_shape = (origin_shape_no, origin_shape_no)
     rgb_index = [0, 1, 2]
     border = (border_no, border_no)
     input_shape = (origin_shape[0] + border[0] + border[1] , origin_shape[1] + border[0] + border[1])
     means, stds = cache_stats(imgs_folder)
     if not path.isdir(pred_folder):mkdir(os.path.join(os.getcwd(),pred_folder))
     if not path.isdir(path.join(pred_folder, model_id)):mkdir(path.join(pred_folder, model_id))
-    if not path.isdir(path.join(pred_folder, model_id)):mkdir(path.join(pred_folder, model_id))
-    if not path.isdir(path.join(path.join(pred_folder, model_name))):mkdir(path.join(path.join(pred_folder, model_id)))
     if model_id == 'resnet_unet':
         model = get_resnet_unet(input_shape, channel_no)
     else:
@@ -64,4 +62,4 @@ def predict(origin_shape, imgs_folder, models_folder, pred_folder, origin_shape_
             mask = mask[mask_index1:mask_index2, mask_index1:mask_index2, ...]
             mask = mask * 255
             mask = mask.astype('uint8')
-            cv2.imwrite(path.join(pred_folder, model_name,'{}.png'.format(img_id)), mask, [cv2.IMWRITE_PNG_COMPRESSION, 9])
+            cv2.imwrite(path.join(pred_folder, model_id,'{}.png'.format(img_id)), mask, [cv2.IMWRITE_PNG_COMPRESSION, 9])

pixel_decoder/train.py

@@ -15,6 +15,7 @@
 from keras import metrics
 from keras.callbacks import ModelCheckpoint
 from pixel_decoder.loss import dice_coef, dice_logloss2, dice_logloss3, dice_coef_rounded, dice_logloss
+from pixel_decoder.resnet_unet import get_resnet_unet
 # import skimage.io
 import keras.backend as K
 
@@ -28,14 +29,13 @@
 # masks_folder = sys.argv[3]
 # models_folder =sys.argv[4]
 
-def train(batch_size, imgs_folder, masks_folder, models_folder, model_id='resnet_unet', origin_shape_no=256, border_no=32, channel_no = 3):
+def train(batch_size, imgs_folder, masks_folder, models_folder, model_id, origin_shape_no, border_no, channel_no):
     origin_shape = (origin_shape_no, origin_shape_no)
     border = (border_no, border_no)
     all_files, all_masks = datafiles(imgs_folder, masks_folder, models_folder)
     means, stds = cache_stats(imgs_folder)
     input_shape = (origin_shape[0] + border[0] + border[1] , origin_shape[1] + border[0] + border[1])
     if model_id == 'resnet_unet':
-        from pixel_decoder.resnet_unet import get_resnet_unet
         model = get_resnet_unet(input_shape, channel_no)
     # elif model_id == 'inception_unet':
     #     from pixel_decoder.inception_unet import get_inception_resnet_v2_unet

pixel_decoder/utils.py

@@ -101,7 +101,7 @@ def rotate_image(image, angle, scale, imgs_folder, masks_folder, models_folder):
 
  # = cache_stats(imgs_folder)
 
-def batch_data_generator(train_idx, batch_size, means, stds, imgs_folder, masks_folder, models_folder, channel_no = 3, border_no=32, origin_shape_no = 256):
+def batch_data_generator(train_idx, batch_size, means, stds, imgs_folder, masks_folder, models_folder, channel_no, border_no, origin_shape_no):
     # all_files, all_masks = datafiles()
     origin_shape = (origin_shape_no, origin_shape_no)
     border = (border_no, border_no)
@@ -160,7 +160,7 @@ def batch_data_generator(train_idx, batch_size, means, stds, imgs_folder, masks_
                 inputs = []
                 outputs = []
 
-def val_data_generator(val_idx, batch_size, validation_steps, means, stds, imgs_folder, masks_folder, models_folder, channel_no = 3, border_no=32, origin_shape_no = 256):
+def val_data_generator(val_idx, batch_size, validation_steps, means, stds, imgs_folder, masks_folder, models_folder, channel_no, border_no, origin_shape_no):
     origin_shape = (origin_shape_no, origin_shape_no)
     border = (border_no, border_no)
     all_files, all_masks = datafiles(imgs_folder, masks_folder, models_folder)

setup.py

@@ -18,7 +18,7 @@
 #readme
 with open('README.md') as f:
     readme = f.read()
-    
+
 setup(
     name='pixel_decoder',
     author='Zhuangfang NaNa Yi',