add pad_to_multiple

research/object_detection/models/faster_rcnn_resnet_v1_fpn_keras_feature_extractor.py

@@ -39,6 +39,7 @@ class ResnetFPN(tf.keras.layers.Layer):
                  backbone_classifier,
                  fpn_features_generator,
                  coarse_feature_layers,
+                 pad_to_multiple,
                  fpn_min_level,
                  resnet_block_names,
                  base_fpn_max_level):
@@ -59,12 +60,13 @@ class ResnetFPN(tf.keras.layers.Layer):
       self.classification_backbone = backbone_classifier
       self.fpn_features_generator = fpn_features_generator
       self.coarse_feature_layers = coarse_feature_layers
+      self.pad_to_multiple = pad_to_multiple
       self._fpn_min_level = fpn_min_level
       self._resnet_block_names = resnet_block_names
       self._base_fpn_max_level = base_fpn_max_level
 
     def call(self, inputs):
-      inputs = ops.pad_to_multiple(inputs, 32)
+      inputs = ops.pad_to_multiple(inputs, self.pad_to_multiple)
       backbone_outputs = self.classification_backbone(inputs)
 
       feature_block_list = []
@@ -102,6 +104,7 @@ class FasterRCNNResnetV1FpnKerasFeatureExtractor(
                first_stage_features_stride,
                conv_hyperparams,
                batch_norm_trainable=False,
+               pad_to_multiple=32,
                weight_decay=0.0,
                fpn_min_level=2,
                fpn_max_level=6,
@@ -152,6 +155,7 @@ class FasterRCNNResnetV1FpnKerasFeatureExtractor(
     self._fpn_max_level = fpn_max_level
     self._additional_layer_depth = additional_layer_depth
     self._freeze_batchnorm = (not batch_norm_trainable)
+    self._pad_to_multiple = pad_to_multiple
     self._override_base_feature_extractor_hyperparams = \
                     override_base_feature_extractor_hyperparams
     self._resnet_block_names = ['block1', 'block2', 'block3', 'block4']
@@ -250,6 +254,7 @@ class FasterRCNNResnetV1FpnKerasFeatureExtractor(
         feature_extractor_model = ResnetFPN(self.classification_backbone,
                            self._fpn_features_generator,
                            self._coarse_feature_layers,
+                           self._pad_to_multiple,
                            self._fpn_min_level,
                            self._resnet_block_names,
                            self._base_fpn_max_level)