add automatic model search for opencv launcher

Author: APrigarina

tools/accuracy_checker/openvino/tools/accuracy_checker/config/config_reader.py

@@ -925,7 +925,7 @@ def provide_precision_and_layout(launchers, input_precisions, input_layouts):
 
 
 def provide_model_type(launcher, arguments):
-    if 'model_type' in arguments:
+    if 'model_type' in arguments and arguments.model_type is not None:
         launcher['_model_type'] = arguments.model_type
     if launcher['framework'] in ['dlsdk', 'openvino', 'g-api'] and 'model_is_blob' in arguments:
         launcher['_model_is_blob'] = arguments.model_is_blob

tools/accuracy_checker/openvino/tools/accuracy_checker/launcher/opencv_launcher.py

@@ -16,13 +16,14 @@
 
 import re
 from collections import OrderedDict
+from pathlib import Path
 import numpy as np
 import cv2
 
 from ..config import PathField, StringField, ConfigError, ListInputsField
 from ..logging import print_info
 from .launcher import Launcher, LauncherConfigValidator
-from ..utils import get_or_parse_value
+from ..utils import get_or_parse_value, get_path
 
 DEVICE_REGEX = r'(?P<device>cpu$|gpu|gpu_fp16)?'
 BACKEND_REGEX = r'(?P<backend>ocv|ie)?'
@@ -63,8 +64,11 @@ class OpenCVLauncher(Launcher):
     def parameters(cls):
         parameters = super().parameters()
         parameters.update({
-            'model': PathField(description="Path to model file."),
-            'weights': PathField(description="Path to weights file.", optional=True, default='', check_exists=False),
+            'model': PathField(description="Path to model file.", file_or_directory=True),
+            'weights': PathField(
+                description="Path to weights file.", optional=True, 
+                check_exists=False, file_or_directory=True
+            ),
             'device': StringField(
                 regex=DEVICE_REGEX, choices=OpenCVLauncher.TARGET_DEVICES.keys(),
                 description="Device name: {}".format(', '.join(OpenCVLauncher.TARGET_DEVICES.keys()))
@@ -100,8 +104,10 @@ def __init__(self, config_entry: dict, *args, **kwargs):
             raise ConfigError('{} is not supported device'.format(selected_device))
 
         if not self._delayed_model_loading:
-            self.model = self.get_value_from_config('model')
-            self.weights = self.get_value_from_config('weights')
+            self.model, self.weights = self.automatic_model_search(self._model_name,
+                self.get_value_from_config('model'), self.get_value_from_config('weights'),
+                self.get_value_from_config('_model_type')
+            )
             self.network = self.create_network(self.model, self.weights)
             self._inputs_shapes = self.get_inputs_from_config(self.config)
             self.network.setInputsNames(list(self._inputs_shapes.keys()))
@@ -130,6 +136,71 @@ def batch(self):
     def output_blob(self):
         return next(iter(self.output_names))
 
+    def automatic_model_search(self, model_name, model_cfg, weights_cfg, model_type=None):
+        model_type_ext = {
+            'xml': 'xml',
+            'blob': 'blob',
+            'onnx': 'onnx',
+            'caffe': 'prototxt',
+            'tf': 'pb'
+        }
+        def get_model_by_suffix(model_name, model_dir, suffix):
+            model_list = list(Path(model_dir).glob('{}.{}'.format(model_name, suffix)))
+            if not model_list:
+                model_list = list(Path(model_dir).glob('*.{}'.format(suffix)))
+            if not model_list:
+                model_list = list(Path(model_dir).parent.rglob('*.{}'.format(suffix)))
+            return model_list
+
+        def get_model():
+            model = Path(model_cfg)
+            if not model.is_dir():
+                accepted_suffixes = list(model_type_ext.values())
+                if model.suffix[1:] not in accepted_suffixes:
+                    raise ConfigError('Models with following suffixes are allowed: {}'.format(accepted_suffixes))
+                print_info('Found model {}'.format(model))
+                return model, model.suffix == '.blob'
+            model_list = []
+            if model_type is not None:
+                model_list = get_model_by_suffix(model_name, model, model_type_ext[model_type])
+            else:
+                for ext in model_type_ext.values():
+                    model_list = get_model_by_suffix(model_name, model, ext)
+                    if model_list:
+                        break
+            if not model_list:
+                raise ConfigError('suitable model is not found')
+            if len(model_list) != 1:
+                raise ConfigError('More than one model matched, please specify explicitly')
+            model = model_list[0]
+            print_info('Found model {}'.format(model))
+            return model, model.suffix == '.blob'
+
+        model, is_blob = get_model()
+        if is_blob:
+            return model, None
+        weights = weights_cfg
+        if (weights is None or Path(weights).is_dir()) and model.suffix != '.onnx':
+            weights_dir = weights or model.parent
+            weights_list = []
+            if model.suffix == '.xml':
+                weights = Path(weights_dir) / model.name.replace('xml', 'bin')
+            else:
+                if model.suffix == '.prototxt':
+                    weights_list = list(Path(weights_dir).glob('*.{}'.format('caffemodel')))
+                if not weights_list:
+                    raise ConfigError('Suitable weights is not detected')
+                if len(weights_list) != 1:
+                    raise ConfigError('Several suitable weights found, please specify required explicitly')
+                weights = weights_list[0]
+        if weights is not None:
+            accepted_weights_suffixes = ['.bin', '.caffemodel']
+            if weights.suffix not in accepted_weights_suffixes:
+                raise ConfigError('Weights with following suffixes are allowed: {}'.format(accepted_weights_suffixes))
+            print_info('Found weights {}'.format(get_path(weights)))
+
+        return model, weights
+
     def predict(self, inputs, metadata=None, **kwargs):
         """
         Args: