fix cpp inference without resize (#1073)

configs/face_detection/blazeface.yml

@@ -96,11 +96,12 @@ EvalReader:
   - !DecodeImage
     to_rgb: true
   - !NormalizeBox {}
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1
 
 TestReader:
@@ -112,9 +113,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

configs/face_detection/blazeface_keypoint.yml

@@ -104,11 +104,12 @@ EvalReader:
   - !DecodeImage
     to_rgb: true
   - !NormalizeBox {}
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1
 
 TestReader:
@@ -120,9 +121,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

configs/face_detection/blazeface_nas.yml

@@ -98,11 +98,12 @@ EvalReader:
   - !DecodeImage
     to_rgb: true
   - !NormalizeBox {}
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1
 
 TestReader:
@@ -114,9 +115,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

configs/face_detection/blazeface_nas_v2.yml

@@ -98,11 +98,12 @@ EvalReader:
   - !DecodeImage
     to_rgb: true
   - !NormalizeBox {}
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1
 
 TestReader:
@@ -114,9 +115,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

configs/face_detection/faceboxes.yml

@@ -97,11 +97,13 @@ EvalReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
+  - !NormalizeBox {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
 
 TestReader:
   inputs_def:
@@ -112,9 +114,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

configs/face_detection/faceboxes_lite.yml

@@ -98,11 +98,13 @@ EvalReader:
   - !DecodeImage
     to_rgb: true
   - !NormalizeBox {}
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
+
 
 TestReader:
   inputs_def:
@@ -113,9 +115,10 @@ TestReader:
   sample_transforms:
   - !DecodeImage
     to_rgb: true
-  - !Permute {}
   - !NormalizeImage
+    is_channel_first: false
     is_scale: false
-    mean: [104, 117, 123]
+    mean: [123, 117, 104]
     std: [127.502231, 127.502231, 127.502231]
+  - !Permute {}
   batch_size: 1

deploy/cpp/include/preprocess_op.h

@@ -50,6 +50,12 @@ class PreprocessOp {
   virtual void Run(cv::Mat* im, ImageBlob* data) = 0;
 };
 
+class InitInfo : public PreprocessOp{
+ public:
+  virtual void Init(const YAML::Node& item, const std::string& arch) {}
+  virtual void Run(cv::Mat* im, ImageBlob* data);
+};
+
 class Normalize : public PreprocessOp {
  public:
   virtual void Init(const YAML::Node& item, const std::string& arch) {
@@ -127,6 +133,8 @@ class Preprocessor {
  public:
   void Init(const YAML::Node& config_node, const std::string& arch) {
     arch_ = arch;
+    // initialize image info at first
+    ops_["InitInfo"] = std::make_shared<InitInfo>();
     for (const auto& item : config_node) {
       auto op_name = item["type"].as<std::string>();
       ops_[op_name] = CreateOp(op_name);

deploy/cpp/src/preprocess_op.cc

@@ -19,6 +19,24 @@
 
 namespace PaddleDetection {
 
+void InitInfo::Run(cv::Mat* im, ImageBlob* data) {
+  data->ori_im_size_ = {
+      static_cast<int>(im->rows),
+      static_cast<int>(im->cols)
+  };
+  data->ori_im_size_f_ = {
+      static_cast<float>(im->rows),
+      static_cast<float>(im->cols),
+      1.0
+  };
+  data->eval_im_size_f_ = {
+    static_cast<float>(im->rows),
+    static_cast<float>(im->cols),
+    1.0
+  };
+  data->scale_factor_f_ = {1., 1., 1., 1.};
+}
+
 void Normalize::Run(cv::Mat* im, ImageBlob* data) {
   double e = 1.0;
   if (is_scale_) {
@@ -44,20 +62,12 @@ void Permute::Run(cv::Mat* im, ImageBlob* data) {
   (data->im_data_).resize(rc * rh * rw);
   float* base = (data->im_data_).data();
   for (int i = 0; i < rc; ++i) {
-    cv::extractChannel(*im, cv::Mat(rh, rw, CV_32FC1, base + i * rh * rw), i);
+    int cur_c = to_bgr_ ? rc - i - 1 : i;
+    cv::extractChannel(*im, cv::Mat(rh, rw, CV_32FC1, base + cur_c * rh * rw), i);
   }
 }
 
 void Resize::Run(cv::Mat* im, ImageBlob* data) {
-  data->ori_im_size_ = {
-      static_cast<int>(im->rows),
-      static_cast<int>(im->cols)
-  };
-  data->ori_im_size_f_ = {
-      static_cast<float>(im->rows),
-      static_cast<float>(im->cols),
-      1.0
-  };
   auto resize_scale = GenerateScale(*im);
   cv::resize(
       *im, *im, cv::Size(), resize_scale.first, resize_scale.second, interp_);
@@ -137,7 +147,7 @@ void PadStride::Run(cv::Mat* im, ImageBlob* data) {
 
 // Preprocessor op running order
 const std::vector<std::string> Preprocessor::RUN_ORDER = {
-  "Resize", "Normalize", "PadStride", "Permute"
+  "InitInfo", "Resize", "Normalize", "PadStride", "Permute"
 };
 
 void Preprocessor::Run(cv::Mat* im, ImageBlob* data) {