initial support for arm_conf/loc structure change

Author: vera121

include/caffe/layers/detection_output_layer.hpp

@@ -40,8 +40,8 @@ class DetectionOutputLayer : public Layer<Dtype> {
 
   virtual inline const char* type() const { return "DetectionOutput"; }
   virtual inline int MinBottomBlobs() const { return 3; }
-  virtual inline int MaxBottomBlobs() const { return 18; }
-  // Note: for no concat cases, the input order is conf*6+loc*6+priorbox(*6)
+  //virtual inline int MaxBottomBlobs() const { return 18; }
+  // Note: for no concat cases, the input order is conf*n+loc*n+priorbox(*n) (+arm_conf*n+arm_loc*n)
   // and only implement for CPU now
   virtual inline int ExactNumTopBlobs() const { return 1; }
 
@@ -118,6 +118,8 @@ class DetectionOutputLayer : public Layer<Dtype> {
   bool conf_concat_ = true;
   bool loc_concat_ = true;
   bool priorbox_concat_ = true;
+  bool arm_conf_no_concat_;
+  bool arm_loc_no_concat_;
 
   bool ratio_permute_;
   bool no_permute_;
@@ -129,11 +131,12 @@ class DetectionOutputLayer : public Layer<Dtype> {
   int ratio5_;
   int nbottom_; // bottom count for conf/loc
   vector<int> collect_ratios_;
-	//TFLite_Detection_Postprocess parameters
-	bool tflite_detection_;
-	bool tflite_use_regular_nms_;
-	vector<float> scale_xywh_;
-	int max_classes_per_detection_;
+
+  //TFLite_Detection_Postprocess parameters
+  bool tflite_detection_;
+  bool tflite_use_regular_nms_;
+  vector<float> scale_xywh_;
+  int max_classes_per_detection_;
 };
 
 }  // namespace caffe

src/caffe/layers/detection_output_layer.cpp

@@ -27,6 +27,8 @@ void DetectionOutputLayer<Dtype>::LayerSetUp(
   ratio3_ = detection_output_param.ratio3();
   ratio4_ = detection_output_param.ratio4();
   ratio5_ = detection_output_param.ratio5();
+  arm_conf_no_concat_ = detection_output_param.arm_conf_no_concat();
+  arm_loc_no_concat_ = detection_output_param.arm_loc_no_concat();
   // TFLite_Detection_PostProcess related params
   tflite_detection_ = detection_output_param.tflite_detection();
   tflite_use_regular_nms_ = detection_output_param.tflite_use_regular_nms();
@@ -139,32 +141,41 @@ void DetectionOutputLayer<Dtype>::LayerSetUp(
     data_transformer_->InitRand();
     save_file_ = detection_output_param.save_file();
   }
-  if (bottom.size() < nbottom_) {
-    conf_concat_ = true;
-    loc_concat_ = true;
-    priorbox_concat_ = true;
-  } else if (bottom.size() >= nbottom_ && bottom.size() < 2 * nbottom_ + 1) {
-    conf_concat_ = false;
-    loc_concat_ = true;
-    priorbox_concat_ = true;
-  } else if ((bottom.size() >= 2 * nbottom_ + 1 &&
-             bottom.size() < 3 * nbottom_) || nbottom_ == 1 ) {
-    conf_concat_ = false;
-    loc_concat_ = false;
-    priorbox_concat_ = true;
-  } else // ==3*nbottom_
+  if (!arm_conf_no_concat_ && !arm_loc_no_concat_)
   {
-    conf_concat_ = false;
-    loc_concat_ = false;
-    priorbox_concat_ = false;
-  }
+    if (bottom.size() < nbottom_) {
+      conf_concat_ = true;
+      loc_concat_ = true;
+      priorbox_concat_ = true;
+    } else if (bottom.size() >= nbottom_ && bottom.size() < 2 * nbottom_ + 1) {
+      conf_concat_ = false;
+      loc_concat_ = true;
+      priorbox_concat_ = true;
+    } else if ((bottom.size() >= 2 * nbottom_ + 1 &&
+               bottom.size() < 3 * nbottom_) || nbottom_ == 1 ) {
+      conf_concat_ = false;
+      loc_concat_ = false;
+      priorbox_concat_ = true;
+    } else // ==3*nbottom_
+    {
+      conf_concat_ = false;
+      loc_concat_ = false;
+      priorbox_concat_ = false;
+    }
 
-  if (conf_concat_ && loc_concat_) {
-    bbox_preds_.ReshapeLike(*(bottom[0]));
-    if (!share_location_) {
-      bbox_permute_.ReshapeLike(*(bottom[0]));
+    if (conf_concat_ && loc_concat_) {
+      bbox_preds_.ReshapeLike(*(bottom[0]));
+      if (!share_location_) {
+        bbox_permute_.ReshapeLike(*(bottom[0]));
+      }
+      conf_permute_.ReshapeLike(*(bottom[1]));
     }
-    conf_permute_.ReshapeLike(*(bottom[1]));
+  }
+  else if (bottom.size() == 4 * nbottom_ + 1) // only consider this situation for mapping with arm_conf/arm_loc
+  {
+    conf_concat_ = false;
+    loc_concat_ = false;
+    priorbox_concat_ = true;
   }
 }
 

src/caffe/proto/caffe.proto

@@ -1698,6 +1698,8 @@ message DetectionOutputParameter {
   optional bool ratio_permute = 25 [default=false];
   optional bool no_permute = 32 [default=false];
   optional int32 nbottom = 33 [default = 6];
+  optional bool arm_conf_no_concat = 38 [default = false];
+  optional bool arm_loc_no_concat = 39 [default = false];
   optional int32 ratio0 = 26 [default = 3];
   optional int32 ratio1 = 27 [default = 6];
   optional int32 ratio2 = 28 [default = 6];