add mkldnn config

deploy/cpp_infer/include/config.h

@@ -39,6 +39,8 @@ public:
     this->cpu_math_library_num_threads =
         stoi(config_map_["cpu_math_library_num_threads"]);
 
+    this->use_mkldnn = bool(stoi(config_map_["use_mkldnn"]));
+
     this->max_side_len = stoi(config_map_["max_side_len"]);
 
     this->det_db_thresh = stod(config_map_["det_db_thresh"]);
@@ -64,6 +66,8 @@ public:
 
   int cpu_math_library_num_threads = 1;
 
+  bool use_mkldnn = false;
+
   int max_side_len = 960;
 
   double det_db_thresh = 0.3;

deploy/cpp_infer/include/ocr_det.h

@@ -39,7 +39,8 @@ public:
   explicit DBDetector(const std::string &model_dir, const bool &use_gpu,
                       const int &gpu_id, const int &gpu_mem,
                       const int &cpu_math_library_num_threads,
-                      const int &max_side_len, const double &det_db_thresh,
+                      const bool &use_mkldnn, const int &max_side_len,
+                      const double &det_db_thresh,
                       const double &det_db_box_thresh,
                       const double &det_db_unclip_ratio,
                       const bool &visualize) {
@@ -47,6 +48,7 @@ public:
     this->gpu_id_ = gpu_id;
     this->gpu_mem_ = gpu_mem;
     this->cpu_math_library_num_threads_ = cpu_math_library_num_threads;
+    this->use_mkldnn_ = use_mkldnn;
 
     this->max_side_len_ = max_side_len;
 
@@ -72,6 +74,7 @@ private:
   int gpu_id_ = 0;
   int gpu_mem_ = 4000;
   int cpu_math_library_num_threads_ = 4;
+  bool use_mkldnn_ = false;
 
   int max_side_len_ = 960;
 

deploy/cpp_infer/include/ocr_rec.h

@@ -38,11 +38,12 @@ public:
   explicit CRNNRecognizer(const std::string &model_dir, const bool &use_gpu,
                           const int &gpu_id, const int &gpu_mem,
                           const int &cpu_math_library_num_threads,
-                          const string &label_path) {
+                          const bool &use_mkldnn, const string &label_path) {
     this->use_gpu_ = use_gpu;
     this->gpu_id_ = gpu_id;
     this->gpu_mem_ = gpu_mem;
     this->cpu_math_library_num_threads_ = cpu_math_library_num_threads;
+    this->use_mkldnn_ = use_mkldnn;
 
     this->label_list_ = Utility::ReadDict(label_path);
 
@@ -61,6 +62,7 @@ private:
   int gpu_id_ = 0;
   int gpu_mem_ = 4000;
   int cpu_math_library_num_threads_ = 4;
+  bool use_mkldnn_ = false;
 
   std::vector<std::string> label_list_;
 

deploy/cpp_infer/src/main.cpp

@@ -50,12 +50,12 @@ int main(int argc, char **argv) {
 
   DBDetector det(config.det_model_dir, config.use_gpu, config.gpu_id,
                  config.gpu_mem, config.cpu_math_library_num_threads,
-                 config.max_side_len, config.det_db_thresh,
+                 config.use_mkldnn, config.max_side_len, config.det_db_thresh,
                  config.det_db_box_thresh, config.det_db_unclip_ratio,
                  config.visualize);
   CRNNRecognizer rec(config.rec_model_dir, config.use_gpu, config.gpu_id,
                      config.gpu_mem, config.cpu_math_library_num_threads,
-                     config.char_list_file);
+                     config.use_mkldnn, config.char_list_file);
 
   auto start = std::chrono::system_clock::now();
   std::vector<std::vector<std::vector<int>>> boxes;

deploy/cpp_infer/src/ocr_det.cpp

@@ -13,7 +13,6 @@
 // limitations under the License.
 
 #include <include/ocr_det.h>
-#include <stdlib.h>
 
 namespace PaddleOCR {
 
@@ -25,7 +24,9 @@ void DBDetector::LoadModel(const std::string &model_dir) {
     config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
   } else {
     config.DisableGpu();
-    // config.EnableMKLDNN(); // not sugesteed to use for now
+    if (this->use_mkldnn_) {
+      config.EnableMKLDNN();
+    }
     config.SetCpuMathLibraryNumThreads(this->cpu_math_library_num_threads_);
   }
 

deploy/cpp_infer/src/ocr_rec.cpp

@@ -12,21 +12,6 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "opencv2/core.hpp"
-#include "opencv2/imgcodecs.hpp"
-#include "opencv2/imgproc.hpp"
-#include "paddle_api.h"
-#include "paddle_inference_api.h"
-#include <chrono>
-#include <iomanip>
-#include <iostream>
-#include <ostream>
-#include <vector>
-
-#include <cstring>
-#include <fstream>
-#include <numeric>
-
 #include <include/ocr_rec.h>
 
 namespace PaddleOCR {
@@ -128,7 +113,9 @@ void CRNNRecognizer::LoadModel(const std::string &model_dir) {
     config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
   } else {
     config.DisableGpu();
-    // config.EnableMKLDNN(); // not sugesteed to use for now
+    if (this->use_mkldnn_) {
+      config.EnableMKLDNN();
+    }
     config.SetCpuMathLibraryNumThreads(this->cpu_math_library_num_threads_);
   }
 

deploy/cpp_infer/tools/config.txt

@@ -3,6 +3,7 @@ use_gpu  0
 gpu_id  0
 gpu_mem  4000
 cpu_math_library_num_threads  10
+use_mkldnn 0
 
 # det config
 max_side_len  960