updating datumaro

cvat/apps/dataset_manager/bindings.py

@@ -1500,7 +1500,7 @@ def get_media_for_frame(self, source_id: int, frame_index: int, **image_kwargs)
         source = self._sources[source_id]
 
         if source.is_video:
-            def video_frame_loader(_):
+            def video_frame_loader():
                 self._load_source(source_id, source)
 
                 # optimization for videos: use numpy arrays instead of bytes
@@ -1510,9 +1510,9 @@ def video_frame_loader(_):
                     out_type=FrameOutputType.NUMPY_ARRAY
                 ).data
 
-            return dm.Image(data=video_frame_loader, **image_kwargs)
+            return dm.Image.from_numpy(data=video_frame_loader, **image_kwargs)
         else:
-            def image_loader(_):
+            def image_loader():
                 self._load_source(source_id, source)
 
                 # for images use encoded data to avoid recoding
@@ -1521,7 +1521,7 @@ def image_loader(_):
                     out_type=FrameOutputType.BUFFER
                 ).data.getvalue()
 
-            return dm.ByteImage(data=image_loader, **image_kwargs)
+            return dm.Image.from_bytes(data=image_loader, **image_kwargs)
 
     def _load_source(self, source_id: int, source: MediaSource) -> None:
         if self._current_source_id == source_id:
@@ -1559,13 +1559,13 @@ def get_media_for_frame(self, source_id: int, frame_id: int, **image_kwargs) ->
         image = self._images_per_source[source_id][frame_id]
 
         related_images = [
-            dm.Image(path=path)
+            dm.Image.from_file(path=path)
             for rf in image.related_files.all()
             for path in [osp.realpath(str(rf.path))]
             if osp.isfile(path)
         ]
 
-        return dm.PointCloud(point_cloud_path, extra_images=related_images)
+        return dm.PointCloud.from_file(point_cloud_path, extra_images=related_images)
 
 MEDIA_PROVIDERS_BY_DIMENSION: dict[DimensionType, MediaProvider] = {
     DimensionType.DIM_3D: MediaProvider3D,
@@ -1689,18 +1689,18 @@ def __init__(
 
                 if not include_images:
                     dm_media_args["extra_images"] = [
-                        dm.Image(path=osp.basename(image.path))
+                        dm.Image.from_file(path=osp.basename(image.path))
                         for image in dm_media.extra_images
                     ]
-                    dm_media = dm.PointCloud(**dm_media_args)
+                    dm_media = dm.PointCloud.from_file(**dm_media_args)
             else:
                 dm_media_args['size'] = (frame_data.height, frame_data.width)
                 if include_images:
                     dm_media: dm.Image = self._media_provider.get_media_for_frame(
                         0, frame_data.idx, **dm_media_args
                     )
                 else:
-                    dm_media = dm.Image(**dm_media_args)
+                    dm_media = dm.Image.from_file(**dm_media_args)
 
             dm_anno = self._read_cvat_anno(frame_data, instance_meta['labels'])
 
@@ -1792,18 +1792,18 @@ def __init__(
 
                 if not include_images:
                     dm_media_args["extra_images"] = [
-                        dm.Image(path=osp.basename(image.path))
+                        dm.Image.from_file(path=osp.basename(image.path))
                         for image in dm_media.extra_images
                     ]
-                    dm_media = dm.PointCloud(**dm_media_args)
+                    dm_media = dm.PointCloud.from_file(**dm_media_args)
             else:
                 dm_media_args['size'] = (frame_data.height, frame_data.width)
                 if include_images:
                     dm_media: dm.Image = self._media_provider.get_media_for_frame(
                         frame_data.task_id, frame_data.idx, **dm_media_args
                     )
                 else:
-                    dm_media = dm.Image(**dm_media_args)
+                    dm_media = dm.Image.from_file(**dm_media_args)
 
             dm_anno = self._read_cvat_anno(frame_data, project_data.meta[project_data.META_FIELD]['labels'])
 

cvat/apps/dataset_manager/formats/camvid.py

@@ -27,7 +27,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         dataset.export(
             temp_dir,
             "camvid",
-            save_images=save_images,
+            save_media=save_images,
             apply_colormap=True,
             label_map={label: label_map[label][0] for label in label_map},
         )

cvat/apps/dataset_manager/formats/cityscapes.py

@@ -33,7 +33,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         dataset.export(
             temp_dir,
             "cityscapes",
-            save_images=save_images,
+            save_media=save_images,
             apply_colormap=True,
             label_map={label: info[0] for label, info in make_colormap(instance_data).items()},
         )

cvat/apps/dataset_manager/formats/coco.py

@@ -24,7 +24,7 @@
 def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "coco_instances", save_images=save_images, merge_images=False)
+        dataset.export(temp_dir, "coco_instances", save_media=save_images, merge_images=False)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -52,7 +52,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         dataset.export(
-            temp_dir, "coco_person_keypoints", save_images=save_images, merge_images=False
+            temp_dir, "coco_person_keypoints", save_media=save_images, merge_images=False
         )
 
     make_zip_archive(temp_dir, dst_file)

cvat/apps/dataset_manager/formats/cvat.py

@@ -115,7 +115,7 @@ def parse_image_dir(image_dir, subset):
                     items[(subset, name)] = DatasetItem(
                         id=name,
                         annotations=[],
-                        media=Image(path=file),
+                        media=Image.from_file(path=file),
                         subset=subset or DEFAULT_SUBSET_NAME,
                     )
 
@@ -617,7 +617,7 @@ def _load_items(self, parsed, image_items):
             )
             image_size = (item_desc.get("height"), item_desc.get("width"))
             if all(image_size):
-                image = Image(path=image, size=tuple(map(int, image_size)))
+                image = Image.from_file(path=image, size=tuple(map(int, image_size)))
             di = image_items.get(
                 (subset, osp.splitext(name)[0]), DatasetItem(id=name, annotations=[])
             )
@@ -1341,7 +1341,7 @@ def load_anno(file_object, annotations):
                         DatasetItem(
                             id=osp.splitext(el.attrib["name"])[0],
                             attributes={"frame": el.attrib["id"]},
-                            media=Image(path=el.attrib["name"]),
+                            media=Image.from_file(path=el.attrib["name"]),
                         ),
                         instance_data=annotations,
                     )

cvat/apps/dataset_manager/formats/datumaro.py

@@ -23,7 +23,7 @@
 def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data=instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "datumaro", save_images=save_images)
+        dataset.export(temp_dir, "datumaro", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -52,7 +52,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         instance_data=instance_data, include_images=save_images, dimension=DimensionType.DIM_3D
     ) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "datumaro", save_images=save_images)
+        dataset.export(temp_dir, "datumaro", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/icdar.py

@@ -78,7 +78,7 @@ def _export_recognition(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         dataset.transform(LabelToCaption)
-        dataset.export(temp_dir, "icdar_word_recognition", save_images=save_images)
+        dataset.export(temp_dir, "icdar_word_recognition", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -101,7 +101,7 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
 def _export_localization(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "icdar_text_localization", save_images=save_images)
+        dataset.export(temp_dir, "icdar_text_localization", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -128,7 +128,7 @@ def _export_segmentation(dst_file, temp_dir, instance_data, save_images=False):
         dataset.transform("polygons_to_masks")
         dataset.transform("boxes_to_masks")
         dataset.transform("merge_instance_segments")
-        dataset.export(temp_dir, "icdar_text_segmentation", save_images=save_images)
+        dataset.export(temp_dir, "icdar_text_segmentation", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/imagenet.py

@@ -8,6 +8,7 @@
 from glob import glob
 
 from datumaro.components.dataset import Dataset
+from datumaro.plugins.transforms import Rename
 
 from cvat.apps.dataset_manager.bindings import GetCVATDataExtractor, import_dm_annotations
 from cvat.apps.dataset_manager.util import make_zip_archive
@@ -20,9 +21,9 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         if save_images:
-            dataset.export(temp_dir, "imagenet", save_images=save_images)
+            dataset.export(temp_dir, "imagenet", save_media=save_images)
         else:
-            dataset.export(temp_dir, "imagenet_txt", save_images=save_images)
+            dataset.export(temp_dir, "imagenet_txt", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -38,6 +39,7 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
         dataset = Dataset.import_from(temp_dir, "imagenet_txt", env=dm_env)
     else:
         dataset = Dataset.import_from(temp_dir, "imagenet", env=dm_env)
+        dataset = dataset.transform(Rename, regex="|([^:]+):(.*)|\\1/\\2|")
         if load_data_callback is not None:
             load_data_callback(dataset, instance_data)
     import_dm_annotations(dataset, instance_data)

cvat/apps/dataset_manager/formats/kitti.py

@@ -34,7 +34,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
             format="kitti",
             label_map={k: v[0] for k, v in make_colormap(instance_data).items()},
             apply_colormap=True,
-            save_images=save_images,
+            save_media=save_images,
         )
 
     make_zip_archive(temp_dir, dst_file)

cvat/apps/dataset_manager/formats/labelme.py

@@ -21,7 +21,7 @@
 def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "label_me", save_images=save_images)
+        dataset.export(temp_dir, "label_me", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/lfw.py

@@ -31,6 +31,6 @@ def _import(src_file, temp_dir, instance_data, load_data_callback=None, **kwargs
 def _exporter(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, format="lfw", save_images=save_images)
+        dataset.export(temp_dir, format="lfw", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)

cvat/apps/dataset_manager/formats/market1501.py

@@ -72,7 +72,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         dataset = Dataset.from_extractors(extractor, env=dm_env)
 
         dataset.transform(LabelAttrToAttr, label="market-1501")
-        dataset.export(temp_dir, "market1501", save_images=save_images)
+        dataset.export(temp_dir, "market1501", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/mask.py

@@ -30,7 +30,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         dataset.export(
             temp_dir,
             "voc_segmentation",
-            save_images=save_images,
+            save_media=save_images,
             apply_colormap=True,
             label_map=make_colormap(instance_data),
         )

cvat/apps/dataset_manager/formats/mot.py

@@ -101,7 +101,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = dm.Dataset.from_extractors(extractor, env=dm_env)
 
-        dataset.export(temp_dir, "mot_seq_gt", save_images=save_images)
+        dataset.export(temp_dir, "mot_seq_gt", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/mots.py

@@ -111,7 +111,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
         dataset.transform("boxes_to_masks")
         dataset.transform("merge_instance_segments")
 
-        dataset.export(temp_dir, "mots_png", save_images=save_images)
+        dataset.export(temp_dir, "mots_png", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/openimages.py

@@ -50,7 +50,7 @@ def _export(dst_file, temp_dir, task_data, save_images=False):
         dataset.transform("polygons_to_masks")
         dataset.transform("merge_instance_segments")
 
-        dataset.export(temp_dir, "open_images", save_images=save_images)
+        dataset.export(temp_dir, "open_images", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/pascal_voc.py

@@ -27,7 +27,7 @@ def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
 
-        dataset.export(temp_dir, "voc", save_images=save_images, label_map="source")
+        dataset.export(temp_dir, "voc", save_media=save_images, label_map="source")
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/pointcloud.py

@@ -28,7 +28,7 @@ def _export_images(dst_file, temp_dir, task_data, save_images=False):
     ) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         dataset.export(
-            temp_dir, "sly_pointcloud", save_images=save_images, allow_undeclared_attrs=True
+            temp_dir, "sly_pointcloud", save_media=save_images, allow_undeclared_attrs=True
         )
 
     make_zip_archive(temp_dir, dst_file)

cvat/apps/dataset_manager/formats/velodynepoint.py

@@ -38,7 +38,7 @@ def _export_images(dst_file, temp_dir, task_data, save_images=False):
     ) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         dataset.transform(RemoveTrackingInformation)
-        dataset.export(temp_dir, "kitti_raw", save_images=save_images, reindex=True)
+        dataset.export(temp_dir, "kitti_raw", save_media=save_images, reindex=True)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/vggface2.py

@@ -22,7 +22,7 @@
 def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "vgg_face2", save_images=save_images)
+        dataset.export(temp_dir, "vgg_face2", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/widerface.py

@@ -21,7 +21,7 @@
 def _export(dst_file, temp_dir, instance_data, save_images=False):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, "wider_face", save_images=save_images)
+        dataset.export(temp_dir, "wider_face", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/dataset_manager/formats/yolo.py

@@ -37,7 +37,7 @@ def _export_common(
 ):
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
-        dataset.export(temp_dir, format_name, save_images=save_images, **kwargs)
+        dataset.export(temp_dir, format_name, save_media=save_images, **kwargs)
 
     make_zip_archive(temp_dir, dst_file)
 
@@ -111,7 +111,7 @@ def _export_yolo_ultralytics_segmentation(dst_file, temp_dir, instance_data, *,
     with GetCVATDataExtractor(instance_data, include_images=save_images) as extractor:
         dataset = Dataset.from_extractors(extractor, env=dm_env)
         dataset = dataset.transform("masks_to_polygons")
-        dataset.export(temp_dir, "yolo_ultralytics_segmentation", save_images=save_images)
+        dataset.export(temp_dir, "yolo_ultralytics_segmentation", save_media=save_images)
 
     make_zip_archive(temp_dir, dst_file)
 

cvat/apps/lambda_manager/views.py

@@ -836,7 +836,9 @@ def parse_anno(anno, labels):
                 elif anno["type"] == "mask":
                     [xtl, ytl, xbr, ybr] = shape["points"][-4:]
                     cut_points = shape["points"][:-4]
-                    rle = mask_tools.mask_to_rle(np.array(cut_points))["counts"].tolist()
+                    rle = mask_tools.mask_to_rle(np.array(cut_points)[:, np.newaxis])[
+                        "counts"
+                    ].tolist()
                     rle.extend([xtl, ytl, xbr, ybr])
                     shape["points"] = rle
 

cvat/requirements/base.in

@@ -12,7 +12,7 @@ azure-storage-blob==12.13.0
 boto3==1.17.61
 clickhouse-connect==0.6.8
 coreapi==2.3.3
-datumaro @ git+https://github.com/cvat-ai/datumaro.git@e61ebe4c5cfee8a741fbf89f966535996be6dcff
+datumaro @ git+https://github.com/cvat-ai/datumaro.git@759c13f43bf86c26fa1e8b9908176e3b902a2203
 dj-pagination==2.5.0
 # Despite direct indication allauth in requirements we should keep 'with_social' for dj-rest-auth
 # to avoid possible further versions conflicts (we use registration functionality)