Python object_detection.builders.box_predictor_builder.build_weight_shared_convolutional_box_predictor() Examples

The following are 30 code examples of object_detection.builders.box_predictor_builder.build_weight_shared_convolutional_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: convolutional_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #2
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #3
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #4
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #5
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #6
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #7
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #8
Source File: convolutional_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #9
Source File: convolutional_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #10
Source File: convolutional_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_bias_predictions_to_background_with_sigmoid_score_conversion(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=True,
              num_classes=2,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              class_prediction_bias_init=-4.6,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      class_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (tf.nn.sigmoid(class_predictions),)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    class_predictions = self.execute(graph_fn, [image_features])
    self.assertAlmostEqual(np.mean(class_predictions), 0.01, places=3)
Example #11
Source File: convolutional_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #12
Source File: convolutional_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 6 votes vote down vote up 
def test_get_boxes_for_five_aspect_ratios_per_location(self):

    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=0,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features], num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      objectness_predictions = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, objectness_predictions)
    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, objectness_predictions) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(objectness_predictions.shape, [4, 320, 1])
Example #13
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_for_five_aspect_ratios_per_location(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 320, num_classes_without_background+1])
Example #14
Source File: convolutional_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_for_five_aspect_ratios_per_location(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 320, num_classes_without_background+1])
Example #15
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_from_two_feature_maps(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2],
          num_predictions_per_location=[5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2])
    self.assertAllEqual(box_encodings.shape, [4, 640, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 640, num_classes_without_background+1])
Example #16
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_from_feature_maps_of_different_depth(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2, image_features3):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2, image_features3],
          num_predictions_per_location=[5, 5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features3 = np.random.rand(4, 8, 8, 32).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2, image_features3])
    self.assertAllEqual(box_encodings.shape, [4, 960, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 960, num_classes_without_background+1])
Example #17
Source File: convolutional_box_predictor_test.py    From g-tensorflow-models with Apache License 2.0 5 votes vote down vote up 
def test_get_predictions_with_feature_maps_of_dynamic_shape(
      self):
    image_features = tf.placeholder(dtype=tf.float32, shape=[4, None, None, 64])
    conv_box_predictor = (
        box_predictor_builder.build_weight_shared_convolutional_box_predictor(
            is_training=False,
            num_classes=0,
            conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
            depth=32,
            num_layers_before_predictor=1,
            box_code_size=4))
    box_predictions = conv_box_predictor.predict(
        [image_features], num_predictions_per_location=[5],
        scope='BoxPredictor')
    box_encodings = tf.concat(box_predictions[box_predictor.BOX_ENCODINGS],
                              axis=1)
    objectness_predictions = tf.concat(box_predictions[
        box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
    init_op = tf.global_variables_initializer()

    resolution = 32
    expected_num_anchors = resolution*resolution*5
    with self.test_session() as sess:
      sess.run(init_op)
      (box_encodings_shape,
       objectness_predictions_shape) = sess.run(
           [tf.shape(box_encodings), tf.shape(objectness_predictions)],
           feed_dict={image_features:
                      np.random.rand(4, resolution, resolution, 64)})
      self.assertAllEqual(box_encodings_shape, [4, expected_num_anchors, 4])
      self.assertAllEqual(objectness_predictions_shape,
                          [4, expected_num_anchors, 1])
Example #18
Source File: convolutional_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_from_feature_maps_of_different_depth(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2, image_features3):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2, image_features3],
          num_predictions_per_location=[5, 5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features3 = np.random.rand(4, 8, 8, 32).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2, image_features3])
    self.assertAllEqual(box_encodings.shape, [4, 960, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 960, num_classes_without_background+1])
Example #19
Source File: convolutional_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_from_two_feature_maps(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2],
          num_predictions_per_location=[5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2])
    self.assertAllEqual(box_encodings.shape, [4, 640, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 640, num_classes_without_background+1])
Example #20
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_for_five_aspect_ratios_per_location(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 320, num_classes_without_background+1])
Example #21
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_from_two_feature_maps(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2],
          num_predictions_per_location=[5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2])
    self.assertAllEqual(box_encodings.shape, [4, 640, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 640, num_classes_without_background+1])
Example #22
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_from_feature_maps_of_different_depth(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2, image_features3):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2, image_features3],
          num_predictions_per_location=[5, 5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features3 = np.random.rand(4, 8, 8, 32).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2, image_features3])
    self.assertAllEqual(box_encodings.shape, [4, 960, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 960, num_classes_without_background+1])
Example #23
Source File: convolutional_box_predictor_tf1_test.py    From models with Apache License 2.0 5 votes vote down vote up 
def test_get_predictions_with_feature_maps_of_dynamic_shape(
      self):
    image_features = tf.placeholder(dtype=tf.float32, shape=[4, None, None, 64])
    conv_box_predictor = (
        box_predictor_builder.build_weight_shared_convolutional_box_predictor(
            is_training=False,
            num_classes=0,
            conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
            depth=32,
            num_layers_before_predictor=1,
            box_code_size=4))
    box_predictions = conv_box_predictor.predict(
        [image_features], num_predictions_per_location=[5],
        scope='BoxPredictor')
    box_encodings = tf.concat(box_predictions[box_predictor.BOX_ENCODINGS],
                              axis=1)
    objectness_predictions = tf.concat(box_predictions[
        box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
    init_op = tf.global_variables_initializer()

    resolution = 32
    expected_num_anchors = resolution*resolution*5
    with self.test_session() as sess:
      sess.run(init_op)
      (box_encodings_shape,
       objectness_predictions_shape) = sess.run(
           [tf.shape(box_encodings), tf.shape(objectness_predictions)],
           feed_dict={image_features:
                      np.random.rand(4, resolution, resolution, 64)})
      self.assertAllEqual(box_encodings_shape, [4, expected_num_anchors, 4])
      self.assertAllEqual(objectness_predictions_shape,
                          [4, expected_num_anchors, 1])
Example #24
Source File: convolutional_box_predictor_test.py    From vehicle_counting_tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_for_five_aspect_ratios_per_location(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 320, num_classes_without_background+1])
Example #25
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_for_five_aspect_ratios_per_location(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features],
          num_predictions_per_location=[5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(box_predictions[
          box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features])
    self.assertAllEqual(box_encodings.shape, [4, 320, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 320, num_classes_without_background+1])
Example #26
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_from_two_feature_maps(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2],
          num_predictions_per_location=[5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2])
    self.assertAllEqual(box_encodings.shape, [4, 640, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 640, num_classes_without_background+1])
Example #27
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 5 votes vote down vote up 
def test_get_multi_class_predictions_from_feature_maps_of_different_depth(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2, image_features3):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2, image_features3],
          num_predictions_per_location=[5, 5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features3 = np.random.rand(4, 8, 8, 32).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2, image_features3])
    self.assertAllEqual(box_encodings.shape, [4, 960, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 960, num_classes_without_background+1])
Example #28
Source File: convolutional_box_predictor_test.py    From multilabel-image-classification-tensorflow with MIT License 5 votes vote down vote up 
def test_get_predictions_with_feature_maps_of_dynamic_shape(
      self):
    image_features = tf.placeholder(dtype=tf.float32, shape=[4, None, None, 64])
    conv_box_predictor = (
        box_predictor_builder.build_weight_shared_convolutional_box_predictor(
            is_training=False,
            num_classes=0,
            conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
            depth=32,
            num_layers_before_predictor=1,
            box_code_size=4))
    box_predictions = conv_box_predictor.predict(
        [image_features], num_predictions_per_location=[5],
        scope='BoxPredictor')
    box_encodings = tf.concat(box_predictions[box_predictor.BOX_ENCODINGS],
                              axis=1)
    objectness_predictions = tf.concat(box_predictions[
        box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
    init_op = tf.global_variables_initializer()

    resolution = 32
    expected_num_anchors = resolution*resolution*5
    with self.test_session() as sess:
      sess.run(init_op)
      (box_encodings_shape,
       objectness_predictions_shape) = sess.run(
           [tf.shape(box_encodings), tf.shape(objectness_predictions)],
           feed_dict={image_features:
                      np.random.rand(4, resolution, resolution, 64)})
      self.assertAllEqual(box_encodings_shape, [4, expected_num_anchors, 4])
      self.assertAllEqual(objectness_predictions_shape,
                          [4, expected_num_anchors, 1])
Example #29
Source File: convolutional_box_predictor_test.py    From BMW-TensorFlow-Training-GUI with Apache License 2.0 5 votes vote down vote up 
def test_get_multi_class_predictions_from_two_feature_maps(
      self):

    num_classes_without_background = 6
    def graph_fn(image_features1, image_features2):
      conv_box_predictor = (
          box_predictor_builder.build_weight_shared_convolutional_box_predictor(
              is_training=False,
              num_classes=num_classes_without_background,
              conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
              depth=32,
              num_layers_before_predictor=1,
              box_code_size=4))
      box_predictions = conv_box_predictor.predict(
          [image_features1, image_features2],
          num_predictions_per_location=[5, 5],
          scope='BoxPredictor')
      box_encodings = tf.concat(
          box_predictions[box_predictor.BOX_ENCODINGS], axis=1)
      class_predictions_with_background = tf.concat(
          box_predictions[box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND],
          axis=1)
      return (box_encodings, class_predictions_with_background)

    image_features1 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    image_features2 = np.random.rand(4, 8, 8, 64).astype(np.float32)
    (box_encodings, class_predictions_with_background) = self.execute(
        graph_fn, [image_features1, image_features2])
    self.assertAllEqual(box_encodings.shape, [4, 640, 4])
    self.assertAllEqual(class_predictions_with_background.shape,
                        [4, 640, num_classes_without_background+1])
Example #30
Source File: convolutional_box_predictor_test.py    From MAX-Object-Detector with Apache License 2.0 5 votes vote down vote up 
def test_get_predictions_with_feature_maps_of_dynamic_shape(
      self):
    image_features = tf.placeholder(dtype=tf.float32, shape=[4, None, None, 64])
    conv_box_predictor = (
        box_predictor_builder.build_weight_shared_convolutional_box_predictor(
            is_training=False,
            num_classes=0,
            conv_hyperparams_fn=self._build_arg_scope_with_conv_hyperparams(),
            depth=32,
            num_layers_before_predictor=1,
            box_code_size=4))
    box_predictions = conv_box_predictor.predict(
        [image_features], num_predictions_per_location=[5],
        scope='BoxPredictor')
    box_encodings = tf.concat(box_predictions[box_predictor.BOX_ENCODINGS],
                              axis=1)
    objectness_predictions = tf.concat(box_predictions[
        box_predictor.CLASS_PREDICTIONS_WITH_BACKGROUND], axis=1)
    init_op = tf.global_variables_initializer()

    resolution = 32
    expected_num_anchors = resolution*resolution*5
    with self.test_session() as sess:
      sess.run(init_op)
      (box_encodings_shape,
       objectness_predictions_shape) = sess.run(
           [tf.shape(box_encodings), tf.shape(objectness_predictions)],
           feed_dict={image_features:
                      np.random.rand(4, resolution, resolution, 64)})
      self.assertAllEqual(box_encodings_shape, [4, expected_num_anchors, 4])
      self.assertAllEqual(objectness_predictions_shape,
                          [4, expected_num_anchors, 1])