Python object_detection.builders.box_predictor_builder.build_mask_rcnn_box_predictor() Examples

The following are 30 code examples of object_detection.builders.box_predictor_builder.build_mask_rcnn_box_predictor(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module object_detection.builders.box_predictor_builder , or try the search function .
Example #1
Source File: mask_rcnn_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #2
Source File: mask_rcnn_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #3
Source File: mask_rcnn_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #4
Source File: mask_rcnn_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #5
Source File: mask_rcnn_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #6
Source File: mask_rcnn_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #7
Source File: mask_rcnn_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #8
Source File: mask_rcnn_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #9
Source File: mask_rcnn_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #10
Source File: mask_rcnn_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #11
Source File: mask_rcnn_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #12
Source File: mask_rcnn_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #13
Source File: mask_rcnn_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #14
Source File: mask_rcnn_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #15
Source File: mask_rcnn_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #16
Source File: mask_rcnn_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #17
Source File: mask_rcnn_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #18
Source File: mask_rcnn_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #19
Source File: mask_rcnn_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #20
Source File: mask_rcnn_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #21
Source File: mask_rcnn_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_with_five_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 5, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #22
Source File: mask_rcnn_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_do_not_return_instance_masks_without_request(self):
    image_features = tf.random_uniform([2, 7, 7, 3], dtype=tf.float32)
    mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
        is_training=False,
        num_classes=5,
        fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
        use_dropout=False,
        dropout_keep_prob=0.5,
        box_code_size=4)
    box_predictions = mask_box_predictor.predict(
        [image_features],
        num_predictions_per_location=[1],
        scope='BoxPredictor',
        prediction_stage=2)
    self.assertEqual(len(box_predictions), 2)
    self.assertTrue(box_predictor.BOX_ENCODINGS in box_predictions)
    self.assertTrue(box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND
                    in box_predictions)
Example #23
Source File: mask_rcnn_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_get_instance_masks(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          conv_hyperparams_fn=self._build_arg_scope_with_hyperparams(
              op_type=hyperparams_pb2.Hyperparams.CONV),
          predict_instance_masks=True)
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=3)
      return (box_predictions[box_predictor.MASK_PREDICTIONS],)
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    mask_predictions = self.execute(graph_fn, [image_features])
    self.assertAllEqual(mask_predictions.shape, [2, 1, 5, 14, 14])
Example #24
Source File: mask_rcnn_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_with_five_classes_share_box_across_classes(self):
    def graph_fn(image_features):
      mask_box_predictor = box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          share_box_across_classes=True
      )
      box_predictions = mask_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[1],
          scope='BoxPredictor',
          prediction_stage=2)
      return (box_predictions[box_predictor.BOX_ENCODINGS],
              box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND])
    image_features = np.random.rand(2, 7, 7, 3).astype(np.float32)
    (box_encodings,
     class_predictions_with_background) = self.execute(graph_fn,
                                                       [image_features])
    self.assertAllEqual(box_encodings.shape, [2, 1, 1, 4])
    self.assertAllEqual(class_predictions_with_background.shape, [2, 1, 6])
Example #25
Source File: mask_rcnn_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)
Example #26
Source File: mask_rcnn_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)
Example #27
Source File: mask_rcnn_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)
Example #28
Source File: mask_rcnn_box_predictor_tf1_test.py    From models with Apache License 2.0 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)
Example #29
Source File: mask_rcnn_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)
Example #30
Source File: mask_rcnn_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 5 votes vote down vote up 
def test_value_error_on_predict_instance_masks_with_no_conv_hyperparms(self):
    with self.assertRaises(ValueError):
      box_predictor_builder.build_mask_rcnn_box_predictor(
          is_training=False,
          num_classes=5,
          fc_hyperparams_fn=self._build_arg_scope_with_hyperparams(),
          use_dropout=False,
          dropout_keep_prob=0.5,
          box_code_size=4,
          predict_instance_masks=True)