Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Fix LFW format #19

Open
wants to merge 3 commits into
base: develop
Choose a base branch
from
Open

Fix LFW format #19

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f70e631
add test
sizov-kirill Dec 1, 2022
1eb4b2e
refactor and fix lfw format
sizov-kirill Dec 1, 2022
6e18272
fix typo
sizov-kirill Dec 2, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
237 changes: 124 additions & 113 deletions datumaro/plugins/lfw_format.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,13 +65,12 @@ def _load_items(self, path):
items = {}
label_categories = self._categories.get(AnnotationType.label)

images = {}
if osp.isdir(self._images_dir):
images = {
osp.splitext(osp.relpath(p, self._images_dir))[0].replace("\\", "/"): p
osp.splitext(osp.relpath(p, self._images_dir))[0].replace("\\", "/"): Image(path=p)
for p in find_images(self._images_dir, recursive=True)
}
else:
images = {}

with open(path, encoding="utf-8") as f:

Expand All @@ -85,51 +84,47 @@ def get_label_id(label_name):

for line in f:
pair = line.strip().split("\t")

if len(pair) == 1 and pair[0] != "":
annotations = []
image = pair[0]
item_id = pair[0]
objects = item_id.split("/")

if 1 < len(objects):
label_name = objects[0]
label = get_label_id(label_name)
if label is not None:
annotations.append(Label(label))
item_id = item_id[len(label_name) + 1 :]
if item_id not in items:
image = images.get(item_id)
if image:
image = Image(path=image)

if item_id not in items:
items[item_id] = DatasetItem(
id=item_id, subset=self._subset, media=image, annotations=annotations
id=item_id,
subset=self._subset,
media=images.get(image),
annotations=annotations,
)

elif len(pair) == 3:
image1, id1 = self.get_image_name(pair[0], pair[1])
image2, id2 = self.get_image_name(pair[0], pair[2])
label = get_label_id(pair[0])

if id1 not in items:
annotations = []
annotations.append(Label(label))

image = images.get(image1)
if image:
image = Image(path=image)

items[id1] = DatasetItem(
id=id1, subset=self._subset, media=image, annotations=annotations
id=id1,
subset=self._subset,
media=images.get(image1),
annotations=[Label(label)],
)
if id2 not in items:
annotations = []
annotations.append(Label(label))

image = images.get(image2)
if image:
image = Image(path=image)

if id2 not in items:
items[id2] = DatasetItem(
id=id2, subset=self._subset, media=image, annotations=annotations
id=id2,
subset=self._subset,
media=images.get(image2),
annotations=[Label(label)],
)

# pairs form a directed graph
Expand All @@ -139,35 +134,32 @@ def get_label_id(label_name):

elif len(pair) == 4:
image1, id1 = self.get_image_name(pair[0], pair[1])

if pair[2] == "-":
image2 = pair[3]
id2 = pair[3]
image2, id2 = pair[3], pair[3]
else:
image2, id2 = self.get_image_name(pair[2], pair[3])

if id1 not in items:
annotations = []
label = get_label_id(pair[0])
annotations.append(Label(label))

image = images.get(image1)
if image:
image = Image(path=image)

items[id1] = DatasetItem(
id=id1, subset=self._subset, media=image, annotations=annotations
id=id1,
subset=self._subset,
media=images.get(image1),
annotations=[Label(label)],
)

if id2 not in items:
annotations = []
if pair[2] != "-":
label = get_label_id(pair[2])
annotations.append(Label(label))

image = images.get(image2)
if image:
image = Image(path=image)

items[id2] = DatasetItem(
id=id2, subset=self._subset, media=image, annotations=annotations
id=id2,
subset=self._subset,
media=images.get(image2),
annotations=annotations,
)

# pairs form a directed graph
Expand All @@ -179,24 +171,35 @@ def get_label_id(label_name):
if osp.isfile(landmarks_file):
with open(landmarks_file, encoding="utf-8") as f:
for line in f:
line = line.split("\t")
line_parts = line.split("\t")
item_id = osp.splitext(line_parts[0])[0]

item_id = osp.splitext(line[0])[0]
objects = item_id.split("/")
if 1 < len(objects):
label_name = objects[0]
label = get_label_id(label_name)
if label is not None:
item_id = item_id[len(label_name) + 1 :]
label_name = objects[0] if 1 < len(objects) else ""

label = get_label_id(label_name)
if label is not None:
item_id = item_id[len(label_name) + 1 :]

if item_id not in items:
items[item_id] = DatasetItem(
id=item_id,
subset=self._subset,
image=osp.join(self._images_dir, line[0]),
image=Image(path=osp.join(self._images_dir, line_parts[0])),
)

annotations = items[item_id].annotations
annotations.append(Points([float(p) for p in line[1:]], label=label))
annotations.append(Points([float(p) for p in line_parts[1:]], label=label))

labeled_images = set(
item.media.path
for item in items.values()
if getattr(item.media, "path", None) is not None
)
for image_name, image in images.items():
if image.path in labeled_images:
continue
items[image_name] = DatasetItem(id=image_name, subset=self._subset, image=image)

return items

Expand Down Expand Up @@ -241,90 +244,94 @@ def apply(self):
label_categories = self._extractor.categories()[AnnotationType.label]
labels = {label.name: 0 for label in label_categories}

unlabeled_items = []
neutral_items = []
included_items = []
positive_pairs = []
negative_pairs = []
neutral_items = []
landmarks = []
included_items = []

for item in subset:
anns = [ann for ann in item.annotations if ann.type == AnnotationType.label]
label, label_name = None, None
if anns:
label = anns[0]
label_name = label_categories[anns[0].label].name
labels[label_name] += 1
label_annotations = [
ann for ann in item.annotations if ann.type == AnnotationType.label
]

if not label_annotations:
unlabeled_items.append(item)
continue

label_obj = label_annotations[0]
label_name = label_categories[label_obj.label].name
labels[label_name] += 1

if self._save_media and item.media:
subdir = osp.join(subset_name, LfwPath.IMAGES_DIR)
if label_name:
subdir = osp.join(subdir, label_name)
subdir = osp.join(subset_name, LfwPath.IMAGES_DIR, label_name)
self._save_image(item, subdir=subdir)

if label is not None:
person1 = label_name
num1 = item.id
if num1.startswith(person1):
num1 = int(num1.replace(person1, "")[1:])
curr_item = person1 + "/" + str(num1)
person1 = label_name
num1 = item.id
if num1.startswith(person1):
num1 = int(num1.replace(person1, "")[1:])
curr_item = person1 + "/" + str(num1)

if "positive_pairs" in label.attributes:
if "positive_pairs" in label_obj.attributes:
if curr_item not in included_items:
included_items.append(curr_item)
for pair in label_obj.attributes["positive_pairs"]:
search = LfwPath.PATTERN.search(pair)
if search:
num2 = search.groups()[1]
num2 = int(num2)
else:
num2 = pair
if num2.startswith(person1):
num2 = num2.replace(person1, "")[1:]
curr_item = person1 + "/" + str(num2)
if curr_item not in included_items:
included_items.append(curr_item)
for pair in label.attributes["positive_pairs"]:
search = LfwPath.PATTERN.search(pair)
if search:
num2 = search.groups()[1]
num2 = int(num2)
else:
num2 = pair
if num2.startswith(person1):
num2 = num2.replace(person1, "")[1:]
curr_item = person1 + "/" + str(num2)
if curr_item not in included_items:
included_items.append(curr_item)
positive_pairs.append("%s\t%s\t%s" % (person1, num1, num2))

if "negative_pairs" in label.attributes:
positive_pairs.append("%s\t%s\t%s" % (person1, num1, num2))

if "negative_pairs" in label_obj.attributes:
if curr_item not in included_items:
included_items.append(curr_item)
for pair in label_obj.attributes["negative_pairs"]:
search = LfwPath.PATTERN.search(pair)
curr_item = ""
if search:
person2, num2 = search.groups()
num2 = int(num2)
curr_item += person2 + "/"
else:
person2 = "-"
num2 = pair
objects = pair.split("/")
if 1 < len(objects) and objects[0] in labels:
person2 = objects[0]
num2 = pair.replace(person2, "")[1:]
curr_item += person2 + "/"
curr_item += str(num2)
if curr_item not in included_items:
included_items.append(curr_item)
for pair in label.attributes["negative_pairs"]:
search = LfwPath.PATTERN.search(pair)
curr_item = ""
if search:
person2, num2 = search.groups()
num2 = int(num2)
curr_item += person2 + "/"
else:
person2 = "-"
num2 = pair
objects = pair.split("/")
if 1 < len(objects) and objects[0] in labels:
person2 = objects[0]
num2 = pair.replace(person2, "")[1:]
curr_item += person2 + "/"
curr_item += str(num2)
if curr_item not in included_items:
included_items.append(curr_item)
negative_pairs.append("%s\t%s\t%s\t%s" % (person1, num1, person2, num2))

if (
"positive_pairs" not in label.attributes
and "negative_pairs" not in label.attributes
and curr_item not in included_items
):
neutral_items.append("%s/%s" % (person1, item.id))
included_items.append(curr_item)
negative_pairs.append("%s\t%s\t%s\t%s" % (person1, num1, person2, num2))

elif item.id not in included_items:
neutral_items.append(item.id)
included_items.append(item.id)
if (
"positive_pairs" not in label_obj.attributes
and "negative_pairs" not in label_obj.attributes
and curr_item not in included_items
):
neutral_items.append("%s/%s" % (person1, item.id))
included_items.append(curr_item)

item_landmarks = [p for p in item.annotations if p.type == AnnotationType.points]
for landmark in item_landmarks:
label_name = label_categories[landmark.label].name
landmarks.append(
"%s\t%s"
% (item.id + LfwPath.IMAGE_EXT, "\t".join(str(p) for p in landmark.points))
"%s/%s\t%s"
% (
label_name,
item.id + LfwPath.IMAGE_EXT,
"\t".join(str(p) for p in landmark.points),
)
)

annotations_dir = osp.join(self._save_dir, subset_name, LfwPath.ANNOTATION_DIR)
Expand All @@ -344,3 +351,7 @@ def apply(self):
people_file = osp.join(annotations_dir, LfwPath.PEOPLE_FILE)
with open(people_file, "w", encoding="utf-8") as f:
f.writelines(["%s\t%d\n" % (label, labels[label]) for label in labels])

if unlabeled_items and self._save_media:
for item in unlabeled_items:
self._save_image(item, subdir=osp.join(item.subset, LfwPath.IMAGES_DIR))
32 changes: 32 additions & 0 deletions tests/test_lfw_format.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,6 +143,38 @@ def test_can_save_and_load_with_landmarks(self):

compare_datasets(self, source_dataset, parsed_dataset)

@mark_requirement(Requirements.DATUM_GENERAL_REQ)
def test_can_save_and_load_with_only_landmarks(self):
source_dataset = Dataset.from_iterable(
[
DatasetItem(
id="name0_0001",
subset="test",
media=Image(data=np.ones((2, 5, 3))),
annotations=[
Points([0, 4, 3, 3, 2, 2, 1, 0, 3, 0], label=0),
],
),
],
categories=["name0"],
)

target_dataset = Dataset.from_iterable(
[
DatasetItem(
id="name0_0001",
subset="test",
media=Image(data=np.ones((2, 5, 3))),
),
],
categories=["name0"],
)
Comment on lines +148 to +171
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this behavior in the test is right. From the format viewpoint, the input annotations are incomplete, and it doesn't seem we can adequately support such cases in the format, so the best approach would be either to complete the missing annotations, if possible, or raise a clear error message otherwise.

My considerations on this specific case with CVAT export are here. Let's try to make the format clearly usable from CVAT, Datumaro seem to be correct at this point.


LfwConverter.convert(source_dataset, "./lfw", save_media=True)
parsed_dataset = Dataset.import_from("./lfw", "lfw")

compare_datasets(self, parsed_dataset, target_dataset)

@mark_requirement(Requirements.DATUM_GENERAL_REQ)
def test_can_save_and_load_with_no_subsets(self):
source_dataset = Dataset.from_iterable(
Expand Down