def get_images_for_image_ids(self, image_ids):
""" returning a list because """
attention_map_file_paths = [
to_image_path_by_id(self.prefix, image_id, file_ending="bbx")
for image_id in image_ids
]
attention_maps = [
load_numpy_from(self.directory_path, image_file_path)
for image_file_path in attention_map_file_paths
]
attention_labels_file_paths = [
to_image_path_by_id(self.prefix, image_id, file_ending="lbx")
for image_id in image_ids
]
attention_labels = [
load_numpy_from(self.directory_path, image_file_path)
for image_file_path in attention_labels_file_paths
]
attention_ids_file_paths = [
to_image_path_by_id(self.prefix, image_id, file_ending="ibx")
for image_id in image_ids
]
attention_ids = [
load_numpy_from(self.directory_path, image_file_path)
for image_file_path in attention_ids_file_paths
]
""" we cannot simply cast to numpy array here b.c. there might be a different amount of attention maps per image """
""" what to do exactly (flatten, concatenate, sum) depends on the caller """
return attention_maps, attention_labels, attention_ids
def load_attention_map_by_image_id_from_many(image_ids, directory_path, image_prefix, split_name=None):
"""
Returns a dict of {image_id : feature_map} for each given image_id.
"""
if split_name:
directory_path = "/".join([directory_path, split_name])
return dict([(image_id, load_numpy_from(to_image_path_by_id(image_prefix, image_id, directory_path, file_ending="bbx")))
for image_id in image_ids])
def load_and_preprocess_single(processable):
file_path = processable["path"]
target_shape = processable["target_shape"]
feature_size = processable["image_feature_size"]
feature_shape = np.array([np.sqrt(feature_size), np.sqrt(feature_size)])
feature_shape = feature_shape.astype("uint8")
image_id = processable["image_id"]
with load_img(file_path) as image:
attention_maps = []
bounding_boxes = processable["boxes"]
for bounding_box in bounding_boxes:
with Image.new(mode="L", size=(image.size)) as bbx:
draw = ImageDraw.Draw(bbx)
draw_box_on_image(draw, bounding_box["box"], fill=True)
#bbx = bbx.resize(target_shape)
bbx = bbx.resize(feature_shape, resample=NEAREST)
attention_map = np.array(bbx)
attention_map = np.reshape(attention_map, feature_size)
attention_maps.append(attention_map)
attention_maps = np.array(attention_maps)
attention_ids = [bounding_box["box_id"] for bounding_box in processable["boxes"]]
attention_ids = np.array(attention_ids)
attention_labels = [bounding_box["category_id"] for bounding_box in processable["boxes"]]
attention_labels = np.array(attention_labels)
directory_path = os.path.dirname(file_path)
image_prefix = processable["image_prefix"]
attention_map_path = to_image_path_by_id(image_prefix, image_id, directory_path, file_ending="bbx")
store_numpy_to(attention_maps, attention_map_path)
attention_label_path = to_image_path_by_id(image_prefix, image_id, directory_path, file_ending="lbx")
store_numpy_to(attention_labels, attention_label_path)
attention_id_path = to_image_path_by_id(image_prefix, image_id, directory_path, file_ending="ibx")
store_numpy_to(attention_ids, attention_id_path)
def get_images_for_image_ids(self, image_ids):
"""
@param captions: the list of captions as dicts of { "caption", "image_id", "id" }
@return: the image in the same order as captions
"""
feature_map_file_paths = [
to_image_path_by_id(self.prefix, image_id, file_ending="npy")
for image_id in image_ids
]
feature_maps = np.array([
load_numpy_from(self.directory_path, image_file_path)
for image_file_path in feature_map_file_paths
])
return feature_maps
def get_images_for_image_ids(self, image_ids):
"""
@param captions: the list of captions as dicts of { "caption", "image_id", "id" }
@return: the image in the same order as captions
"""
image_file_paths = [
to_image_path_by_id(self.prefix, image_id, self.directory_path)
for image_id in image_ids
]
images = np.array([
_get_image(image_file_path, self.target_size)
for image_file_path in image_file_paths
])
if self.vgg_like:
images = preprocess_input(images, mode="caffe")
return images
def get_images_for_image_ids(self, image_ids):
"""
@param captions: the list of captions as dicts of { "caption", "image_id", "id" }
@return: the image in the same order as captions
"""
image_file_paths = [
to_image_path_by_id(self.prefix, image_id, self.directory_path)
for image_id in image_ids
]
""" original sized images b.c. boxes are from there """
images = [
get_image(image_file_path) for image_file_path in image_file_paths
]
image_boxes = self.get_boxes_by_image_ids(image_ids)
images = draw_boxes_on_images(images, image_boxes)
images = resize_images(images, self.target_size)
return images, image_boxes
def __store_feature_maps_to_many(image_ids,
feature_maps,
directory_path,
image_prefix,
split_name=None):
"""
@param ids: array 1d (batches)
A list of image ids in the same order as the feature maps
@param maps: array 3d (batches, features, maps)
An array of feature maps in the same order as the ids
"""
if split_name:
directory_path = "/".join([directory_path, split_name])
return [
store_numpy_to(
feature_map,
to_image_path_by_id(image_prefix,
image_id,
directory_path,
file_ending="npy"))
for image_id, feature_map in zip(image_ids, feature_maps)
]
def _get_image_for_image_id(self, image_id):
""" Overwritten to determine the file ending """
image_file_path = to_image_path_by_id(self.prefix,
image_id,
file_ending="bbx")
return load_numpy_from(self.directory_path, image_file_path)