Support for VersaDataset (#3270)

Author: mzuevx

tools/accuracy_checker/openvino/tools/accuracy_checker/annotation_converters/README.md

@@ -832,6 +832,11 @@ The main difference between this converter and `super_resolution` in data organi
   * `images_dir` - path to directory with images (optional, default image_2).
   * `label_start` - specifies label index start in label map. Optional, default value is 1. You can provide another value, if you want to use this dataset for separate label validation.
   * `images_suffix` - suffix for image file names (Optional, default: `.png`).
+* `malware_classification_converter` - converts dataset for malware detection task to `ClassificationAnnotation`.
+  * `annotation_file` - path to dataset annotation file.
+  * `data_dir` - path to dataset root folder.
+  * `known_folder` - name of folder containing known files (optional, default `KNOWN_1000`).
+  * `malicious_folder` - name of folder containing malicious files (optional, default `MALICIOUS_1000`).
 
 ## <a name="customizing-dataset-meta"></a>Customizing Dataset Meta
 There are situations when we need to customize some default dataset parameters (e.g. replace original dataset label map with own.)

tools/accuracy_checker/openvino/tools/accuracy_checker/annotation_converters/__init__.py

@@ -129,6 +129,7 @@
 from .gan_annotation_converter import GANAnnotationConverter
 from .kitti_converter import KITTIConverter
 from .smartlab_action_recognition import SmartLabActionRecognition
+from .malware_classification import MalwareClassificationDatasetConverter
 
 __all__ = [
     'BaseFormatConverter',
@@ -258,4 +259,5 @@
     'GANAnnotationConverter',
     'KITTIConverter',
     'SmartLabActionRecognition',
+    'MalwareClassificationDatasetConverter',
 ]

tools/accuracy_checker/openvino/tools/accuracy_checker/annotation_converters/malware_classification.py

@@ -0,0 +1,67 @@
+"""
+Copyright (c) 2018-2022 Intel Corporation
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+      http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+"""
+
+from ..representation import ClassificationAnnotation
+from ..utils import get_path, read_csv, check_file_existence
+from ..config import StringField, PathField
+from .format_converter import FileBasedAnnotationConverter
+from .format_converter import ConverterReturn
+
+class MalwareClassificationDatasetConverter(FileBasedAnnotationConverter):
+    __provider__ = 'malware_classification_converter'
+
+    @classmethod
+    def parameters(cls):
+        parameters = super().parameters()
+        parameters.update({
+            'known_folder': StringField(optional=True, default='KNOWN_1000',
+                                        description="Known files folder"),
+            'malicious_folder': StringField(optional=True, default='MALICIOUS_1000',
+                                            description="Malicious files folder"),
+            'data_dir': PathField(optional=False, default=None, is_directory=True, description="Dataset root folder"),
+        })
+
+        return parameters
+
+    def configure(self):
+        self.annotation_file = self.get_value_from_config('annotation_file')
+        self.data_folder = self.get_value_from_config('data_dir')
+        self.known_folder = self.get_value_from_config('known_folder')
+        self.malicious_folder = self.get_value_from_config('malicious_folder')
+
+        self.data_dir = get_path(self.data_folder, is_directory=True)
+        self.known_dir = get_path(self.data_folder / self.known_folder, is_directory=True)
+        self.malicious_dir = get_path(self.data_folder / self.malicious_folder, is_directory=True)
+
+    def convert(self, check_content=False, progress_callback=None, progress_interval=100, **kwargs):
+        annotations = []
+        content_errors = []
+        meta = None
+
+        annotation = read_csv(self.annotation_file, is_dict=False)
+
+        for entry_id, (identifier, kind) in enumerate(annotation):
+            identifier = identifier.replace(':', '_')
+            kind = int(kind)
+            sample_path = self.malicious_dir / identifier if kind == 1 else self.known_dir / identifier
+            if check_content:
+                if not check_file_existence(sample_path):
+                    content_errors.append('{}: does not exist'.format(sample_path))
+            annotations.append(ClassificationAnnotation(str(sample_path.relative_to(self.data_dir)), [kind, ]))
+            if progress_callback is not None and entry_id % progress_interval == 0:
+                progress_callback(entry_id * 100 / len(annotation))
+
+        return ConverterReturn(annotations, meta, content_errors)