Skip to content

GitLab

Commit 68c9be95 authored by Julian Dierkes's avatar Julian Dierkes
Browse files

added preprocessing to CNNSupervisedTrainer and fixed tests

parent 252d858e
Pipeline #223970 failed with stages
in 50 seconds
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 1761 additions and 538 deletions
src/main/resources/templates/gluon/CNNDataLoader.ftl
View file @ 68c9be95
......@@ -65,48 +65,6 @@ class ${tc.fileNameWithoutEnding}:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -148,6 +106,9 @@ class ${tc.fileNameWithoutEnding}:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -178,11 +139,14 @@ class ${tc.fileNameWithoutEnding}:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/main/resources/templates/gluon/CNNGanTrainer.ftl
View file @ 68c9be95
......@@ -114,7 +114,7 @@ class ${tc.fileNameWithoutEnding}:
# if preprocessing:
# train_iter, test_iter, data_mean, data_std = self._data_loader.load_preprocessed_data(batch_size, preproc_lib)
# else:
# train_iter, test_iter, data_mean, data_std = self._data_loader.load_data(batch_size, img_resize)
# train_iter, test_iter, data_mean, data_std = self._data_loader.load_data(batch_size)
if 'weight_decay' in optimizer_params:
optimizer_params['wd'] = optimizer_params['weight_decay']
......@@ -166,16 +166,6 @@ class ${tc.fileNameWithoutEnding}:
loss_function = mx.gluon.loss.SigmoidBinaryCrossEntropyLoss()
activation_name = 'sigmoid'
<<<<<<< HEAD
=======
<#list tc.architecture.streams as stream>
input_shape = <#list tc.getStreamInputDimensions(stream) as dimensions>${tc.join(dimensions, ",")}</#list>
</#list>
shape_list = list(input_shape)
shape_list[0] = batch_size
input_shape = tuple(shape_list)
>>>>>>> cdd27689fde2eae0cb6e022d18400f7a696b2f58
metric_dis = mx.metric.create(eval_metric)
metric_gen = mx.metric.create(eval_metric)
<#include "gan/InputGenerator.ftl">
......
src/main/resources/templates/gluon/CNNSupervisedTrainer.ftl
View file @ 68c9be95
......@@ -192,7 +192,8 @@ class ${tc.fileNameWithoutEnding}:
context='gpu',
save_attention_image=False,
use_teacher_forcing=False,
normalize=True):
normalize=True,
preprocessing = False):
if context == 'gpu':
mx_context = mx.gpu()
elif context == 'cpu':
......@@ -200,6 +201,12 @@ class ${tc.fileNameWithoutEnding}:
else:
logging.error("Context argument is '" + context + "'. Only 'cpu' and 'gpu are valid arguments'.")
if preprocessing:
preproc_lib = "CNNPreprocessor_${tc.fileNameWithoutEnding?keep_after("CNNSupervisedTrainer_")}_executor"
train_iter, test_iter, data_mean, data_std, train_images, test_images = self._data_loader.load_preprocessed_data(batch_size, preproc_lib)
else:
train_iter, test_iter, data_mean, data_std, train_images, test_images = self._data_loader.load_data(batch_size)
if 'weight_decay' in optimizer_params:
optimizer_params['wd'] = optimizer_params['weight_decay']
del optimizer_params['weight_decay']
......@@ -215,8 +222,6 @@ class ${tc.fileNameWithoutEnding}:
del optimizer_params['step_size']
del optimizer_params['learning_rate_decay']
train_iter, test_iter, data_mean, data_std, train_images, test_images = self._data_loader.load_data(batch_size)
if normalize:
self._net_creator.construct(context=mx_context, data_mean=data_mean, data_std=data_std)
else:
......
src/main/resources/templates/gluon/CNNTrainer.ftl
View file @ 68c9be95
......@@ -43,6 +43,9 @@ if __name__ == "__main__":
<#if (config.normalize)??>
normalize=${config.normalize?string("True","False")},
</#if>
<#if (config.preprocessingName)??>
preprocessing=${config.preprocessingName???string("True","False")},
</#if>
<#if (config.useTeacherForcing)??>
use_teacher_forcing='${config.useTeacherForcing?string("True","False")}',
</#if>
......
src/test/resources/target_code/CNNDataLoader_Alexnet.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_Alexnet:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -147,6 +105,9 @@ class CNNDataLoader_Alexnet:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -177,11 +138,14 @@ class CNNDataLoader_Alexnet:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/test/resources/target_code/CNNDataLoader_CifarClassifierNetwork.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_CifarClassifierNetwork:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -147,6 +105,9 @@ class CNNDataLoader_CifarClassifierNetwork:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -177,11 +138,14 @@ class CNNDataLoader_CifarClassifierNetwork:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/test/resources/target_code/CNNDataLoader_Invariant.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_Invariant:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -147,6 +105,9 @@ class CNNDataLoader_Invariant:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -177,11 +138,14 @@ class CNNDataLoader_Invariant:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/test/resources/target_code/CNNDataLoader_MultipleStreams.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_MultipleStreams:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -147,6 +105,9 @@ class CNNDataLoader_MultipleStreams:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -177,11 +138,14 @@ class CNNDataLoader_MultipleStreams:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/test/resources/target_code/CNNDataLoader_RNNencdec.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_RNNencdec:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()
......@@ -147,6 +105,9 @@ class CNNDataLoader_RNNencdec:
data_mean[input_name + '_'] = nd.array(train_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(train_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
if 'images' in train_h5:
train_images = train_h5['images']
train_iter = mx.io.NDArrayIter(data=train_data,
label=train_label,
batch_size=batch_size)
......@@ -177,11 +138,14 @@ class CNNDataLoader_RNNencdec:
for output_name in self._output_names_:
test_label[output_name][i] = getattr(shape_output, output_name + "_out")
if 'images' in test_h5:
test_images = test_h5['images']
test_iter = mx.io.NDArrayIter(data=test_data,
label=test_label,
batch_size=batch_size)
return train_iter, test_iter, data_mean, data_std
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def preprocess_data(self, instance_wrapper, input_wrapper, index, data_h5):
for input_name in self._input_names_:
......
src/test/resources/target_code/CNNDataLoader_RNNsearch.py
View file @ 68c9be95
......@@ -64,48 +64,6 @@ class CNNDataLoader_RNNsearch:
return train_iter, test_iter, data_mean, data_std, train_images, test_images
def load_data_img(self, batch_size, img_size):
train_h5, test_h5 = self.load_h5_files()
width = img_size[0]
height = img_size[1]
comb_data = {}
data_mean = {}
data_std = {}
for input_name in self._input_names_:
train_data = train_h5[input_name][:]
test_data = test_h5[input_name][:]
train_shape = train_data.shape
test_shape = test_data.shape
comb_data[input_name] = mx.nd.zeros((train_shape[0]+test_shape[0], train_shape[1], width, height))
for i, img in enumerate(train_data):
img = img.transpose(1,2,0)
comb_data[input_name][i] = cv2.resize(img, (width, height)).reshape((train_shape[1],width,height))
for i, img in enumerate(test_data):
img = img.transpose(1, 2, 0)
comb_data[input_name][i+train_shape[0]] = cv2.resize(img, (width, height)).reshape((train_shape[1], width, height))
data_mean[input_name + '_'] = nd.array(comb_data[input_name][:].mean(axis=0))
data_std[input_name + '_'] = nd.array(comb_data[input_name][:].asnumpy().std(axis=0) + 1e-5)
comb_label = {}
for output_name in self._output_names_:
train_labels = train_h5[output_name][:]
test_labels = test_h5[output_name][:]
comb_label[output_name] = np.append(train_labels, test_labels, axis=0)
train_iter = mx.io.NDArrayIter(data=comb_data,
label=comb_label,
batch_size=batch_size)
test_iter = None
return train_iter, test_iter, data_mean, data_std
def load_preprocessed_data(self, batch_size, preproc_lib):
train_h5, test_h5 = self.load_h5_files()