added task import feature for audino

cvat/apps/dataset_manager/formats/LibriVox.py

@@ -0,0 +1,104 @@
+import os.path as osp
+import zipfile
+from glob import glob
+from cvat.apps.dataset_manager.bindings import InstanceLabelData
+from cvat.apps.engine.serializers import LabeledDataSerializer
+import cvat.apps.dataset_manager as dm
+from cvat.apps.dataset_manager.task import PatchAction
+from .registry import importer
+
+
+
+
+
+
+
+
+def load_anno(file_object, annotations):
+    if isinstance(file_object, str):
+        with open(file_object, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+    lines = content.splitlines()
+    headers = lines[0].split('\t')
+
+    label_data = InstanceLabelData(annotations.db_instance)
+
+
+
+
+    for line in lines[1:]:
+        fields = line.split('\t')
+        record = dict(zip(headers, fields))
+
+        job_id = record.get('job_id')
+
+        start = float(record.get('start', 0))
+        end = float(record.get('end', 0))
+
+        label_name = record.get('label')
+        label_id = label_data._get_label_id(label_name)
+
+        language_id_to_locale_mapping = {0: "en"}
+        language_id = int(record.get('language',0))
+
+        spec_id = label_data._get_attribute_id(label_id,record.get("attribute_1_name"))
+
+
+
+        shapes_data = [
+            {
+                "type": "rectangle",
+                "label": record.get("label", ""),
+                "points": [start, start, end, end],
+                "frame":0,
+                "occluded" : False,
+                "z_order": 0,
+                "group": None,
+                "source": "manual",
+                "transcript": record.get("text", ""),
+                "gender": record.get("gender", ""),
+                "age": record.get("age",""),
+                "locale":language_id_to_locale_mapping.get(language_id, ""),
+                "accent": record.get("accent",""),
+                "emotion": record.get("emotion", ""),
+                "rotation": 0.0,
+                "label_id": label_id,
+                "attributes": [
+            {
+                "spec_id": spec_id,
+                "value": record.get("attribute_1_value", ""),
+            } ]
+            }
+        ]
+
+
+        data = {
+            'shapes': shapes_data
+        }
+
+        serializer = LabeledDataSerializer(data=data)
+        pk = int(job_id)
+        action = PatchAction.CREATE
+
+        if serializer.is_valid(raise_exception=True):
+                data = dm.task.patch_job_data(pk, serializer.data, action)
+
+
+
+
+
+
+
+
+
+@importer(name='LibriVox', ext='TSV, ZIP', version=" ")
+def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
+    is_zip = zipfile.is_zipfile(src_file)
+    src_file.seek(0)
+    if is_zip:
+        zipfile.ZipFile(src_file).extractall(temp_dir)
+
+        anno_paths = glob(osp.join(temp_dir, '**', '*.tsv'), recursive=True)
+        for p in anno_paths:
+            load_anno(p, instance_data)

cvat/apps/dataset_manager/formats/VCTK_Corpus.py

@@ -0,0 +1,105 @@
+import os.path as osp
+import zipfile
+from glob import glob
+from cvat.apps.dataset_manager.bindings import InstanceLabelData
+from cvat.apps.engine.serializers import LabeledDataSerializer
+import cvat.apps.dataset_manager as dm
+from cvat.apps.dataset_manager.task import PatchAction
+from .registry import  importer
+
+
+
+
+
+
+
+
+def load_anno(file_object, annotations):
+    if isinstance(file_object, str):
+        with open(file_object, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+    lines = content.splitlines()
+    headers = lines[0].split('\t')
+
+    label_data = InstanceLabelData(annotations.db_instance)
+
+
+
+
+    for line in lines[1:]:
+        fields = line.split('\t')
+        record = dict(zip(headers, fields))
+
+        job_id = record.get('job_id')
+
+        start = float(record.get('start', 0))
+        end = float(record.get('end', 0))
+
+        label_name = record.get('label')
+        label_id = label_data._get_label_id(label_name)
+
+        language_id_to_locale_mapping = {0: "en"}
+        language_id = int(record.get('language',0))
+
+
+        spec_id = label_data._get_attribute_id(label_id,record.get("attribute_1_name"))
+
+
+
+        shapes_data = [
+            {
+                "type": "rectangle",
+                "label": record.get("label", ""),
+                "points": [start, start, end, end],
+                "frame":0,
+                "occluded" : False,
+                "z_order": 0,
+                "group": None,
+                "source": "manual",
+                "transcript": record.get("text", ""),
+                "gender": record.get("gender", ""),
+                "age": record.get("age",""),
+                "locale":language_id_to_locale_mapping.get(language_id, ""),
+                "accent": record.get("accent",""),
+                "emotion": record.get("emotion", ""),
+                "rotation": 0.0,
+                "label_id": label_id,
+                "attributes": [
+            {
+                "spec_id": spec_id,
+                "value": record.get("attribute_1_value", ""),
+            } ]
+            }
+        ]
+
+
+        data = {
+            'shapes': shapes_data
+        }
+
+        serializer = LabeledDataSerializer(data=data)
+        pk = int(job_id)
+        action = PatchAction.CREATE
+
+        if serializer.is_valid(raise_exception=True):
+                data = dm.task.patch_job_data(pk, serializer.data, action)
+
+
+
+
+
+
+
+
+
+@importer(name='VCTK Corpus', ext='TSV, ZIP', version=" ")
+def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
+    is_zip = zipfile.is_zipfile(src_file)
+    src_file.seek(0)
+    if is_zip:
+        zipfile.ZipFile(src_file).extractall(temp_dir)
+
+        anno_paths = glob(osp.join(temp_dir, '**', '*.tsv'), recursive=True)
+        for p in anno_paths:
+            load_anno(p, instance_data)

cvat/apps/dataset_manager/formats/VoxCeleb.py

@@ -0,0 +1,105 @@
+import os.path as osp
+import zipfile
+from glob import glob
+from cvat.apps.dataset_manager.bindings import InstanceLabelData
+from cvat.apps.engine.serializers import LabeledDataSerializer
+import cvat.apps.dataset_manager as dm
+from cvat.apps.dataset_manager.task import PatchAction
+from .registry import importer
+
+
+
+
+
+
+
+
+def load_anno(file_object, annotations):
+    if isinstance(file_object, str):
+        with open(file_object, 'r', encoding='utf-8') as f:
+            content = f.read()
+
+    lines = content.splitlines()
+    headers = lines[0].split('\t')
+
+    label_data = InstanceLabelData(annotations.db_instance)
+
+
+
+
+    for line in lines[1:]:
+        fields = line.split('\t')
+        record = dict(zip(headers, fields))
+
+        job_id = record.get('job_id')
+
+        start = float(record.get('start', 0))
+        end = float(record.get('end', 0))
+
+        label_name = record.get('label')
+        label_id = label_data._get_label_id(label_name)
+
+        language_id_to_locale_mapping = {0: "en"}
+        language_id = int(record.get('language',0))
+
+
+        spec_id = label_data._get_attribute_id(label_id,record.get("attribute_1_name"))
+
+
+
+        shapes_data = [
+            {
+                "type": "rectangle",
+                "label": record.get("label", ""),
+                "points": [start, start, end, end],
+                "frame":0,
+                "occluded" : False,
+                "z_order": 0,
+                "group": None,
+                "source": "manual",
+                "transcript": record.get("text", ""),
+                "gender": record.get("gender", ""),
+                "age": record.get("age",""),
+                "locale":language_id_to_locale_mapping.get(language_id, ""),
+                "accent": record.get("accent",""),
+                "emotion": record.get("emotion", ""),
+                "rotation": 0.0,
+                "label_id": label_id,
+                "attributes": [
+            {
+                "spec_id": spec_id,
+                "value": record.get("attribute_1_value", ""),
+            } ]
+            }
+        ]
+
+
+        data = {
+            'shapes': shapes_data
+        }
+
+        serializer = LabeledDataSerializer(data=data)
+        pk = int(job_id)
+        action = PatchAction.CREATE
+
+        if serializer.is_valid(raise_exception=True):
+                data = dm.task.patch_job_data(pk, serializer.data, action)
+
+
+
+
+
+
+
+
+
+@importer(name='VoxCeleb', ext='TSV, ZIP', version=" ")
+def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs):
+    is_zip = zipfile.is_zipfile(src_file)
+    src_file.seek(0)
+    if is_zip:
+        zipfile.ZipFile(src_file).extractall(temp_dir)
+
+        anno_paths = glob(osp.join(temp_dir, '**', '*.tsv'), recursive=True)
+        for p in anno_paths:
+            load_anno(p, instance_data)