def test_disallow_duplicate_keypoint_ids(self):
label_map_str = """
item {
id: 1
name: 'person'
keypoints: {
id: 1
label: 'right_elbow'
}
keypoints: {
id: 1
label: 'left_elbow'
}
}
item {
id: 2
name: 'face'
keypoints: {
id: 3
label: 'ear'
}
}
"""
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
text_format.Merge(label_map_str, label_map_proto)
with self.assertRaises(ValueError):
label_map_util.convert_label_map_to_categories(label_map_proto,
max_num_classes=2)
def main(tfrecords_filename, label_map=None):
"""
Visualize the image and label stored in tf record
Blatantly stole from
https://stackoverflow.com/questions/50391967/how-to-visualize-a-tfrecord
:param tfrecords_filename: full path to the tfrecord file
:param label_map: None by default
:return:
"""
matplotlib.use('TkAgg')
if label_map is not None:
label_map_proto = pb.StringIntLabelMap()
with tf.gfile.GFile(label_map, 'r') as f:
text_format.Merge(f.read(), label_map_proto)
class_dict = {}
for entry in label_map_proto.item:
class_dict[entry.id] = {'name': entry.display_name}
sess = tf.Session()
decoder = TfDecoder(label_map_proto_file=label_map, use_display_name=False)
sess.run(tf.tables_initializer())
for record in tf.python_io.tf_record_iterator(tfrecords_filename):
example = decoder.decode(record)
host_example = sess.run(example)
scores = np.ones(host_example['groundtruth_boxes'].shape[0])
vu.visualize_boxes_and_labels_on_image_array(
host_example['image'],
host_example['groundtruth_boxes'],
host_example['groundtruth_classes'],
scores,
class_dict,
max_boxes_to_draw=None,
use_normalized_coordinates=True)
plt.imshow(host_example['image'])
plt.show()
def load_labelmap(labelmap_path):
"""Loads labelmap from the labelmap path.
Args:
labelmap_path: Path to the labelmap.
Returns:
A dictionary mapping class name to class numerical id
A list with dictionaries, one dictionary per category.
"""
# pylint: disable=no-member
label_map = string_int_label_map_pb2.StringIntLabelMap()
with open(labelmap_path, "r") as fid:
label_map_string = fid.read()
text_format.Merge(label_map_string, label_map)
labelmap_dict = {}
categories = []
for item in label_map.item:
labelmap_dict[item.name] = item.id
categories.append({
"id": item.id,
"name": item.name,
"display_name": item.display_name,
})
return labelmap_dict, categories
def test_keep_categories_with_unique_id(self):
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
label_map_string = """
item {
id:2
name:'cat'
}
item {
id:1
name:'child'
}
item {
id:1
name:'person'
}
item {
id:1
name:'n00007846'
}
"""
text_format.Merge(label_map_string, label_map_proto)
categories = label_map_util.convert_label_map_to_categories(
label_map_proto, max_num_classes=3)
self.assertListEqual([{
'id': 2,
'name': u'cat'
}, {
'id': 1,
'name': u'child'
}], categories)
def load_labelmap(path):
"""Loads label map proto.
Args:
path: path to StringIntLabelMap proto text file.
Returns:
a StringIntLabelMapProto
"""
df = pd.read_csv(path)
label_map = string_int_label_map_pb2.StringIntLabelMap()
records = df.to_dict('records')
def label_map_func(attr):
o = OrderedDict(attr[1])
v = list(o.values())
a = v[0].split(" ")
a = list(filter(lambda x: len(x.strip()) > 0, a))
label_map_item = string_int_label_map_pb2.StringIntLabelMapItem()
if len(attr) >= 1 and len(a) == 2 and int(a[1]) > 0:
label_map_item.name = str(a[0])
label_map_item.id = int(a[1])
label_map.item.append(label_map_item)
return 1
else:
return False
for ii, record in enumerate(records):
label_map_func((ii, record))
print("Validating now..., Records: ", len(label_map.item))
_validate_label_map(label_map)
return label_map
def _generate_label_map(self, num_classes):
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
for i in range(1, num_classes + 1):
item = label_map_proto.item.add()
item.id = i
item.name = 'label_' + str(i)
item.display_name = str(i)
return label_map_proto
def generate_label_map(labels):
num_classes = len(labels)
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
for i in range(num_classes):
item = label_map_proto.item.add()
item.id = i + 1
item.name = labels[i]
item.display_name = ' '.join(labels[i].split('_'))
return str(label_map_proto)
def class_dict_to_label_map_str(class_dict):
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
for key,val in class_dict.items():
item = label_map_proto.item.add()
item.name = key
# 0 is reserved for 'background' only, which we aren't using
item.id = val + 1
return text_format.MessageToString(label_map_proto)
def load_labelmap(path):
with tf.gfile.GFile(path, 'r') as fid:
label_map_string = fid.read()
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
_validate_label_map(label_map)
return label_map
def save_label_map(self):
label_map = string_int_label_map_pb2.StringIntLabelMap()
for kkk, vvv in self.label_map_dict.items():
new_item = string_int_label_map_pb2.StringIntLabelMapItem()
new_item.name = vvv['name']
new_item.id = vvv['id']
label_map.item.append(new_item)
lb_string = text_format.MessageToString(label_map)
label_map_path = os.path.join(self.input_dir_path, "label_map.pbtxt")
with tf.gfile.Open(label_map_path, "wb") as f:
f.write(lb_string)
print("Label map saved: {0}".format(label_map_path))
def _generate_label_map_with_hierarchy(self, num_classes, ancestors_dict,
descendants_dict):
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
for i in range(1, num_classes + 1):
item = label_map_proto.item.add()
item.id = i
item.name = 'label_' + str(i)
item.display_name = str(i)
if i in ancestors_dict:
for anc_i in ancestors_dict[i]:
item.ancestor_ids.append(anc_i)
if i in descendants_dict:
for desc_i in descendants_dict[i]:
item.descendant_ids.append(desc_i)
return label_map_proto
def write_label_map(categories, label_map_path): label_map = string_int_label_map_pb2.StringIntLabelMap() label_map_items = [] for i, category in enumerate(categories): idx = i + 1 proto = string_int_label_map_pb2.StringIntLabelMapItem() proto.id = idx proto.name = category label_map_items.append(proto) label_map.item.extend(label_map_items) label_map_str = text_format.MessageToString(label_map) with open(label_map_path, 'w') as label_map_file: label_map_file.write(label_map_str)
def load_classes(self, label_map_path, use_display_name=False):
with open(label_map_path) as f:
label_map_string = f.read()
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
label_map_dict = {}
for item in label_map.item:
if use_display_name:
label_map_dict[item.display_name] = item.id
else:
label_map_dict[item.name] = item.id
return label_map_dict
def load_labelmap(path):
"""Loads label map proto.
Args:
path: path to StringIntLabelMap proto text file.
Returns:
a StringIntLabelMapProto
"""
with tf.gfile.GFile(path, "r") as fid:
label_map_string = fid.read()
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
return label_map
def load_labelmap(path):
"""Loads label map proto.
Args:
path: path to StringIntLabelMap proto text file.
Returns:
a StringIntLabelMapProto
"""
label_map_string = read_content(path)
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
_validate_label_map(label_map)
return label_map
def test_get_label_map_dict_from_proto(self):
label_map_string = """
item {
id:2
name:'cat'
}
item {
id:1
name:'dog'
}
"""
label_map_proto = text_format.Parse(
label_map_string, string_int_label_map_pb2.StringIntLabelMap())
label_map_dict = label_map_util.get_label_map_dict(label_map_proto)
self.assertEqual(label_map_dict['dog'], 1)
self.assertEqual(label_map_dict['cat'], 2)
def get_label_map_from_cvat_tfrecord_zip(cvat_tf_record_zip):
"""Extract label map from cvat tfrecord zip file.
CVAT's tfrecord file contains:
- label_map.pbtxt
- *.tfrecord
"""
with tempfile.TemporaryDirectory() as temp_dir:
with ZipFile(cvat_tf_record_zip) as cur_zip:
with cur_zip.open('label_map.pbtxt', 'r') as f:
content = f.read().decode('utf-8')
labels = []
cur_label_map = string_int_label_map_pb2.StringIntLabelMap()
text_format.Merge(content, cur_label_map)
for item in cur_label_map.item:
if item.name not in labels:
labels.append(item.name)
return labels
def load_label_map(path):
"""
:param path: Path to StringIntLabelMap proto text file
:return: A StringIntLabelMapProto
"""
with tf.gfile.GFile(path, 'r') as fid:
label_map_string = fid.read()
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
validate_label_map(label_map)
return label_map
def load_labelmap(path):
"""Loads label map proto.
Args:
path: path to StringIntLabelMap proto text file.
Returns:
a StringIntLabelMapProto
"""
with tf.gfile.GFile(path, 'r') as fid:
#label_map_string = fid.read().decode() ---> Used with TF 1.0
label_map_string = fid.read()
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(label_map_string, label_map)
except text_format.ParseError:
label_map.ParseFromString(label_map_string)
_validate_label_map(label_map)
return label_map
def _load_labelmap(labelmap_path):
"""Loads labelmap from the labelmap path.
Args:
labelmap_path: Path to the labelmap.
Returns:
A dictionary mapping class name to class numerical id.
"""
label_map = string_int_label_map_pb2.StringIntLabelMap()
with open(labelmap_path, 'r') as fid:
label_map_string = fid.read()
text_format.Merge(label_map_string, label_map)
labelmap_dict = {}
for item in label_map.item:
labelmap_dict[item.name] = item.id
return labelmap_dict
def test_convert_label_map_to_categories_lvis_frequency_and_counts(self):
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
label_map_string = """
item {
id:1
name:'person'
frequency: FREQUENT
instance_count: 1000
}
item {
id:2
name:'dog'
frequency: COMMON
instance_count: 100
}
item {
id:3
name:'cat'
frequency: RARE
instance_count: 10
}
"""
text_format.Parse(label_map_string, label_map_proto)
categories = label_map_util.convert_label_map_to_categories(
label_map_proto, max_num_classes=3)
self.assertListEqual([{
'id': 1,
'name': u'person',
'frequency': 'f',
'instance_count': 1000
}, {
'id': 2,
'name': u'dog',
'frequency': 'c',
'instance_count': 100
}, {
'id': 3,
'name': u'cat',
'frequency': 'r',
'instance_count': 10
}], categories)
def detection():
data = request.get_data()
json_data = json.loads(data.decode("utf-8"))
flawCategorys = json_data["flawCategorys"]
taskFileSet = json_data["taskFileSet"]
label_map_string = label_string(flawCategorys)
label_map = string_int_label_map_pb2.StringIntLabelMap()
label_map = text_format.Merge(label_map_string, label_map)
categories = label_map_util.convert_label_map_to_categories(
label_map, max_num_classes=len(flawCategorys), use_display_name=True)
category_index = label_map_util.create_category_index(categories)
threading.Thread(target=detection_power,
args=(
taskFileSet,
category_index,
detection_graph,
configs,
)).start()
status = {'status': 0}
return jsonify(status)
def _load_labelmap(labelmap_path):
"""Loads labelmap from the labelmap path.
Args:
labelmap_path: Path to the labelmap.
Returns:
A dictionary mapping class name to class numerical id
A list with dictionaries, one dictionary per category.
"""
label_map = string_int_label_map_pb2.StringIntLabelMap()
with open(labelmap_path, 'r') as fid:
label_map_string = fid.read()
text_format.Merge(label_map_string, label_map)
labelmap_dict = {}
categories = []
for item in label_map.item:
labelmap_dict[item.name] = item.id
categories.append({'id': item.id, 'name': item.name})
return labelmap_dict, categories
def draw_tfrecord(tfrecords_filename, label_map=None):
"""
Draw the image and bounding box indicated in the groundtruth tf record file.
:param tfrecords_filename:
:param label_map:
:return:
"""
if label_map is not None:
label_map_proto = pb.StringIntLabelMap()
with tf.gfile.GFile(label_map, 'r') as f:
text_format.Merge(f.read(), label_map_proto)
class_dict = {}
for entry in label_map_proto.item:
class_dict[entry.id] = {'name': entry.display_name}
sess = tf.Session()
decoder = TfDecoder(label_map_proto_file=label_map, use_display_name=False)
sess.run(tf.tables_initializer())
iter_num = 0
for record in tf.python_io.tf_record_iterator(tfrecords_filename):
iter_num += 1
print("This is the {} example".format(iter_num))
example = decoder.decode(record)
host_example = sess.run(example)
# Set the score of the groundth truth bounding box as 1
scores = np.ones(host_example['groundtruth_boxes'].shape[0])
vu.visualize_boxes_and_labels_on_image_array(
host_example['image'],
host_example['groundtruth_boxes'],
host_example['groundtruth_classes'],
scores,
class_dict,
max_boxes_to_draw=None,
use_normalized_coordinates=True)
plt.imshow(host_example['image'])
plt.show(block=False)
plt.pause(.1)
plt.close("all")
def make(filename):
"""Make labels and save into a file"""
label_map = string_int_label_map_pb2.StringIntLabelMap()
all_items = []
res = api.get_classes()
if isinstance(res, dict) and 'data' in res.keys():
data = res['data']
total = res['total']
if total == 0:
all_items.append('person') # No class then use person only
for _c in data:
all_items.append(_c['class'])
for i in range(len(all_items)):
item = label_map.item.add()
item.id = i + 1
item.name = all_items[i]
print("{}: {}".format(i + 1, all_items[i]))
f = open(filename, "w")
f.write(text_format.MessageToString(label_map))
f.close()
def test_convert_label_map_with_keypoints_to_categories(self):
label_map_str = """
item {
id: 1
name: 'person'
keypoints: {
id: 1
label: 'nose'
}
keypoints: {
id: 2
label: 'ear'
}
}
"""
label_map_proto = string_int_label_map_pb2.StringIntLabelMap()
text_format.Merge(label_map_str, label_map_proto)
categories = label_map_util.convert_label_map_to_categories(
label_map_proto, max_num_classes=1)
self.assertEqual('person', categories[0]['name'])
self.assertEqual(1, categories[0]['id'])
self.assertEqual(1, categories[0]['keypoints']['nose'])
self.assertEqual(2, categories[0]['keypoints']['ear'])
#%%
from google.protobuf import text_format
from object_detection.protos import string_int_label_map_pb2
#%%
with open('./res/labelmap.pbtxt', "rt") as fd:
# fd = open('./res/labelmap.pbtxt',"rt")
_text = fd.read()
print(_text)
# %%
# 파싱
label_map = string_int_label_map_pb2.StringIntLabelMap()
try:
text_format.Merge(_text, label_map)
except text_format.ParseError:
label_map.ParseFromString(_text)
# print(label_map)
_dic = {}
for item in label_map.item:
print(item)
print(item.name)
print(item.id)
_dic[item.name] = item.id
print(_dic)
# %%
from google.protobuf import text_format from object_detection.protos import string_int_label_map_pb2 LABEL_MAP = "data/label_map.pb.txt" f = open(LABEL_MAP).read() m = text_format.Parse(f, string_int_label_map_pb2.StringIntLabelMap()) layers = [item.name for item in m.item] layer_dict = dict([(item.id, item.name) for item in m.item])
