Merge pull request #8894 from syiming:adjust_proto_file_frcnn_fpn

PiperOrigin-RevId: 324505842

research/object_detection/protos/faster_rcnn.proto

@@ -8,6 +8,7 @@ import "object_detection/protos/hyperparams.proto";
 import "object_detection/protos/image_resizer.proto";
 import "object_detection/protos/losses.proto";
 import "object_detection/protos/post_processing.proto";
+import "object_detection/protos/fpn.proto";
 
 // Configuration for Faster R-CNN models.
 // See meta_architectures/faster_rcnn_meta_arch.py and models/model_builder.py
@@ -212,4 +213,21 @@ message FasterRcnnFeatureExtractor {
   // When training with a relative large batch size (e.g. 8), it could be
   // desirable to enable batch norm update.
   optional bool batch_norm_trainable = 3 [default = false];
+
+  // Hyperparameters that affect the layers of feature extractor added on top
+  // of the base feature extractor.
+  optional Hyperparams conv_hyperparams = 4;
+
+  // if the value is set to true, the base feature extractor's hyperparams will
+  // be overridden with the `conv_hyperparams`.
+  optional bool override_base_feature_extractor_hyperparams = 5
+      [default = false];
+
+  // The nearest multiple to zero-pad the input height and width dimensions to.
+  // For example, if pad_to_multiple = 2, input dimensions are zero-padded
+  // until the resulting dimensions are even.
+  optional int32 pad_to_multiple = 6 [default = 32];
+
+  // Feature Pyramid Networks config.
+  optional FeaturePyramidNetworks fpn = 7;
 }

research/object_detection/protos/fpn.proto

+syntax = "proto2";
+
+package object_detection.protos;
+
+// Configuration for Feature Pyramid Networks.
+message FeaturePyramidNetworks {
+  // We recommend to use multi_resolution_feature_map_generator with FPN, and
+  // the levels there must match the levels defined below for better
+  // performance.
+  // Correspondence from FPN levels to Resnet/Mobilenet V1 feature maps:
+  // FPN Level        Resnet Feature Map      Mobilenet-V1 Feature Map
+  //     2               Block 1                Conv2d_3_pointwise
+  //     3               Block 2                Conv2d_5_pointwise
+  //     4               Block 3                Conv2d_11_pointwise
+  //     5               Block 4                Conv2d_13_pointwise
+  //     6               Bottomup_5             bottom_up_Conv2d_14
+  //     7               Bottomup_6             bottom_up_Conv2d_15
+  //     8               Bottomup_7             bottom_up_Conv2d_16
+  //     9               Bottomup_8             bottom_up_Conv2d_17
+
+  // minimum level in feature pyramid
+  optional int32 min_level = 1 [default = 3];
+
+  // maximum level in feature pyramid
+  optional int32 max_level = 2 [default = 7];
+
+  // channel depth for additional coarse feature layers.
+  optional int32 additional_layer_depth = 3 [default = 256];
+
+}
+
+// Configuration for Bidirectional Feature Pyramid Networks.
+message BidirectionalFeaturePyramidNetworks {
+  // minimum level in the feature pyramid.
+  optional int32 min_level = 1 [default = 3];
+
+  // maximum level in the feature pyramid.
+  optional int32 max_level = 2 [default = 7];
+
+  // The number of repeated top-down bottom-up iterations for BiFPN-based
+  // feature extractors (bidirectional feature pyramid networks).
+  optional int32 num_iterations = 3;
+
+  // The number of filters (channels) to use in feature pyramid layers for
+  // BiFPN-based feature extractors (bidirectional feature pyramid networks).
+  optional int32 num_filters = 4;
+
+  // Method used to combine inputs to BiFPN nodes.
+  optional string combine_method = 5 [default = 'fast_attention'];
+}

research/object_detection/protos/ssd.proto

@@ -11,6 +11,7 @@ import "object_detection/protos/losses.proto";
 import "object_detection/protos/matcher.proto";
 import "object_detection/protos/post_processing.proto";
 import "object_detection/protos/region_similarity_calculator.proto";
+import "object_detection/protos/fpn.proto";
 
 // Configuration for Single Shot Detection (SSD) models.
 // Next id: 27
@@ -203,50 +204,3 @@ message SsdFeatureExtractor {
 
 }
 
-// Configuration for Feature Pyramid Networks.
-message FeaturePyramidNetworks {
-  // We recommend to use multi_resolution_feature_map_generator with FPN, and
-  // the levels there must match the levels defined below for better
-  // performance.
-  // Correspondence from FPN levels to Resnet/Mobilenet V1 feature maps:
-  // FPN Level        Resnet Feature Map      Mobilenet-V1 Feature Map
-  //     2               Block 1                Conv2d_3_pointwise
-  //     3               Block 2                Conv2d_5_pointwise
-  //     4               Block 3                Conv2d_11_pointwise
-  //     5               Block 4                Conv2d_13_pointwise
-  //     6               Bottomup_5             bottom_up_Conv2d_14
-  //     7               Bottomup_6             bottom_up_Conv2d_15
-  //     8               Bottomup_7             bottom_up_Conv2d_16
-  //     9               Bottomup_8             bottom_up_Conv2d_17
-
-  // minimum level in feature pyramid
-  optional int32 min_level = 1 [default = 3];
-
-  // maximum level in feature pyramid
-  optional int32 max_level = 2 [default = 7];
-
-  // channel depth for additional coarse feature layers.
-  optional int32 additional_layer_depth = 3 [default = 256];
-
-}
-
-// Configuration for Bidirectional Feature Pyramid Networks.
-message BidirectionalFeaturePyramidNetworks {
-  // minimum level in the feature pyramid.
-  optional int32 min_level = 1 [default = 3];
-
-  // maximum level in the feature pyramid.
-  optional int32 max_level = 2 [default = 7];
-
-  // The number of repeated top-down bottom-up iterations for BiFPN-based
-  // feature extractors (bidirectional feature pyramid networks).
-  optional int32 num_iterations = 3;
-
-  // The number of filters (channels) to use in feature pyramid layers for
-  // BiFPN-based feature extractors (bidirectional feature pyramid networks).
-  optional int32 num_filters = 4;
-
-  // Method used to combine inputs to BiFPN nodes.
-  optional string combine_method = 5 [default = 'fast_attention'];
-}
-