2martens/masterthesis
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Disable augmentation of input for now

Browse Source 
Signed-off-by: Jim Martens <github@2martens.de>
This commit is contained in:
Jim Martens
2019-07-02 15:10:40 +02:00
parent e4e94940b8
commit 112dc48f36
2 changed files with 8 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/twomartens/masterthesis/cli.py
View File

@ -75,13 +75,13 @@ def _ssd_train(args: argparse.Namespace) -> None:
predictor_sizes=ssd_model.predictor_sizes,
batch_size=batch_size,
resized_shape=(image_size, image_size),
training=True, evaluation=False)
training=True, evaluation=False, augment=False)
val_generator, val_length = \
data.load_scenenet_data(file_names_val, instances_val, args.coco_path,
predictor_sizes=ssd_model.predictor_sizes,
batch_size=batch_size,
resized_shape=(image_size, image_size),
training=False, evaluation=False)
training=False, evaluation=False, augment=False)
del file_names_train, instances_train, file_names_val, instances_val
if args.debug:

10
src/twomartens/masterthesis/data.py
View File

@ -238,7 +238,8 @@ def load_scenenet_data(photo_paths: Sequence[Sequence[str]],
batch_size: int,
resized_shape: Sequence[int],
training: bool,
evaluation: bool) -> Tuple[callable, int]:
evaluation: bool,
augment: bool) -> Tuple[callable, int]:
"""
Loads the SceneNet RGB-D data and returns a data set.
@ -251,6 +252,7 @@ def load_scenenet_data(photo_paths: Sequence[Sequence[str]],
resized_shape: shape of input images to SSD
training: True if training data is desired
evaluation: True if evaluation-ready data is desired
augment: True if training data should be augmented
Returns:
scenenet data set generator
@ -296,14 +298,14 @@ def load_scenenet_data(photo_paths: Sequence[Sequence[str]],
labels=final_labels
)
if training:
shuffle = True
shuffle = True if training else False
if training and augment:
transformations = [data_augmentation_chain_original_ssd.SSDDataAugmentation(
img_width=resized_shape[0],
img_height=resized_shape[1]
)]
else:
shuffle = False
transformations = [
object_detection_2d_photometric_ops.ConvertTo3Channels(),
object_detection_2d_geometric_ops.Resize(height=resized_shape[0],