merge upstream

deploy/cpp_infer/include/utility.h

@@ -44,6 +44,9 @@ public:
   inline static size_t argmax(ForwardIterator first, ForwardIterator last) {
     return std::distance(first, std::max_element(first, last));
   }
+
+  static void GetAllFiles(const char *dir_name,
+                          std::vector<std::string> &all_inputs);
 };
 
 } // namespace PaddleOCR
\ No newline at end of file

deploy/cpp_infer/src/main.cpp

@@ -27,9 +27,12 @@
 #include <fstream>
 #include <numeric>
 
+#include <glog/logging.h>
 #include <include/config.h>
 #include <include/ocr_det.h>
 #include <include/ocr_rec.h>
+#include <include/utility.h>
+#include <sys/stat.h>
 
 using namespace std;
 using namespace cv;
@@ -47,13 +50,8 @@ int main(int argc, char **argv) {
   config.PrintConfigInfo();
 
   std::string img_path(argv[2]);
-
-  cv::Mat srcimg = cv::imread(img_path, cv::IMREAD_COLOR);
-
-  if (!srcimg.data) {
-    std::cerr << "[ERROR] image read failed! image path: " << img_path << "\n";
-    exit(1);
-  }
+  std::vector<std::string> all_img_names;
+  Utility::GetAllFiles((char *)img_path.c_str(), all_img_names);
 
   DBDetector det(config.det_model_dir, config.use_gpu, config.gpu_id,
                  config.gpu_mem, config.cpu_math_library_num_threads,
@@ -76,18 +74,30 @@ int main(int argc, char **argv) {
                      config.use_tensorrt, config.use_fp16);
 
   auto start = std::chrono::system_clock::now();
-  std::vector<std::vector<std::vector<int>>> boxes;
-  det.Run(srcimg, boxes);
-
-  rec.Run(boxes, srcimg, cls);
-  auto end = std::chrono::system_clock::now();
-  auto duration =
-      std::chrono::duration_cast<std::chrono::microseconds>(end - start);
-  std::cout << "Cost  "
-            << double(duration.count()) *
-                   std::chrono::microseconds::period::num /
-                   std::chrono::microseconds::period::den
-            << "s" << std::endl;
+
+  for (auto img_dir : all_img_names) {
+    LOG(INFO) << "The predict img: " << img_dir;
+
+    cv::Mat srcimg = cv::imread(img_dir, cv::IMREAD_COLOR);
+    if (!srcimg.data) {
+      std::cerr << "[ERROR] image read failed! image path: " << img_path
+                << "\n";
+      exit(1);
+    }
+    std::vector<std::vector<std::vector<int>>> boxes;
+
+    det.Run(srcimg, boxes);
+
+    rec.Run(boxes, srcimg, cls);
+    auto end = std::chrono::system_clock::now();
+    auto duration =
+        std::chrono::duration_cast<std::chrono::microseconds>(end - start);
+    std::cout << "Cost  "
+              << double(duration.count()) *
+                     std::chrono::microseconds::period::num /
+                     std::chrono::microseconds::period::den
+              << "s" << std::endl;
+  }
 
   return 0;
 }

deploy/cpp_infer/src/ocr_det.cpp

@@ -30,6 +30,42 @@ void DBDetector::LoadModel(const std::string &model_dir) {
           this->use_fp16_ ? paddle_infer::Config::Precision::kHalf
                           : paddle_infer::Config::Precision::kFloat32,
           false, false);
+      std::map<std::string, std::vector<int>> min_input_shape = {
+          {"x", {1, 3, 50, 50}},
+          {"conv2d_92.tmp_0", {1, 96, 20, 20}},
+          {"conv2d_91.tmp_0", {1, 96, 10, 10}},
+          {"nearest_interp_v2_1.tmp_0", {1, 96, 10, 10}},
+          {"nearest_interp_v2_2.tmp_0", {1, 96, 20, 20}},
+          {"nearest_interp_v2_3.tmp_0", {1, 24, 20, 20}},
+          {"nearest_interp_v2_4.tmp_0", {1, 24, 20, 20}},
+          {"nearest_interp_v2_5.tmp_0", {1, 24, 20, 20}},
+          {"elementwise_add_7", {1, 56, 2, 2}},
+          {"nearest_interp_v2_0.tmp_0", {1, 96, 2, 2}}};
+      std::map<std::string, std::vector<int>> max_input_shape = {
+          {"x", {1, 3, this->max_side_len_, this->max_side_len_}},
+          {"conv2d_92.tmp_0", {1, 96, 400, 400}},
+          {"conv2d_91.tmp_0", {1, 96, 200, 200}},
+          {"nearest_interp_v2_1.tmp_0", {1, 96, 200, 200}},
+          {"nearest_interp_v2_2.tmp_0", {1, 96, 400, 400}},
+          {"nearest_interp_v2_3.tmp_0", {1, 24, 400, 400}},
+          {"nearest_interp_v2_4.tmp_0", {1, 24, 400, 400}},
+          {"nearest_interp_v2_5.tmp_0", {1, 24, 400, 400}},
+          {"elementwise_add_7", {1, 56, 400, 400}},
+          {"nearest_interp_v2_0.tmp_0", {1, 96, 400, 400}}};
+      std::map<std::string, std::vector<int>> opt_input_shape = {
+          {"x", {1, 3, 640, 640}},
+          {"conv2d_92.tmp_0", {1, 96, 160, 160}},
+          {"conv2d_91.tmp_0", {1, 96, 80, 80}},
+          {"nearest_interp_v2_1.tmp_0", {1, 96, 80, 80}},
+          {"nearest_interp_v2_2.tmp_0", {1, 96, 160, 160}},
+          {"nearest_interp_v2_3.tmp_0", {1, 24, 160, 160}},
+          {"nearest_interp_v2_4.tmp_0", {1, 24, 160, 160}},
+          {"nearest_interp_v2_5.tmp_0", {1, 24, 160, 160}},
+          {"elementwise_add_7", {1, 56, 40, 40}},
+          {"nearest_interp_v2_0.tmp_0", {1, 96, 40, 40}}};
+
+      config.SetTRTDynamicShapeInfo(min_input_shape, max_input_shape,
+                                    opt_input_shape);
     }
   } else {
     config.DisableGpu();
@@ -48,7 +84,7 @@ void DBDetector::LoadModel(const std::string &model_dir) {
   config.SwitchIrOptim(true);
 
   config.EnableMemoryOptim();
-  config.DisableGlogInfo();
+  // config.DisableGlogInfo();
 
   this->predictor_ = CreatePredictor(config);
 }

deploy/cpp_infer/src/ocr_rec.cpp

@@ -25,8 +25,9 @@ void CRNNRecognizer::Run(std::vector<std::vector<std::vector<int>>> boxes,
 
   std::cout << "The predicted text is :" << std::endl;
   int index = 0;
-  for (int i = boxes.size() - 1; i >= 0; i--) {
+  for (int i = 0; i < boxes.size(); i++) {
     crop_img = GetRotateCropImage(srcimg, boxes[i]);
+
     if (cls != nullptr) {
       crop_img = cls->Run(crop_img);
     }
@@ -105,6 +106,15 @@ void CRNNRecognizer::LoadModel(const std::string &model_dir) {
           this->use_fp16_ ? paddle_infer::Config::Precision::kHalf
                           : paddle_infer::Config::Precision::kFloat32,
           false, false);
+      std::map<std::string, std::vector<int>> min_input_shape = {
+          {"x", {1, 3, 32, 10}}};
+      std::map<std::string, std::vector<int>> max_input_shape = {
+          {"x", {1, 3, 32, 2000}}};
+      std::map<std::string, std::vector<int>> opt_input_shape = {
+          {"x", {1, 3, 32, 320}}};
+
+      config.SetTRTDynamicShapeInfo(min_input_shape, max_input_shape,
+                                    opt_input_shape);
     }
   } else {
     config.DisableGpu();

deploy/cpp_infer/src/preprocess_op.cpp

@@ -77,19 +77,13 @@ void ResizeImgType0::Run(const cv::Mat &img, cv::Mat &resize_img,
 
   int resize_h = int(float(h) * ratio);
   int resize_w = int(float(w) * ratio);
-  
+
   resize_h = max(int(round(float(resize_h) / 32) * 32), 32);
   resize_w = max(int(round(float(resize_w) / 32) * 32), 32);
 
-  if (!use_tensorrt) {
-    cv::resize(img, resize_img, cv::Size(resize_w, resize_h));
-    ratio_h = float(resize_h) / float(h);
-    ratio_w = float(resize_w) / float(w);
-  } else {
-    cv::resize(img, resize_img, cv::Size(640, 640));
-    ratio_h = float(640) / float(h);
-    ratio_w = float(640) / float(w);
-  }
+  cv::resize(img, resize_img, cv::Size(resize_w, resize_h));
+  ratio_h = float(resize_h) / float(h);
+  ratio_w = float(resize_w) / float(w);
 }
 
 void CrnnResizeImg::Run(const cv::Mat &img, cv::Mat &resize_img, float wh_ratio,
@@ -108,23 +102,12 @@ void CrnnResizeImg::Run(const cv::Mat &img, cv::Mat &resize_img, float wh_ratio,
     resize_w = imgW;
   else
     resize_w = int(ceilf(imgH * ratio));
-  if (!use_tensorrt) {
-    cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
-               cv::INTER_LINEAR);
-    cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0,
-                       int(imgW - resize_img.cols), cv::BORDER_CONSTANT,
-                       {127, 127, 127});
-  } else {
-    int k = int(img.cols * 32 / img.rows);
-    if (k >= 100) {
-      cv::resize(img, resize_img, cv::Size(100, 32), 0.f, 0.f,
-                 cv::INTER_LINEAR);
-    } else {
-      cv::resize(img, resize_img, cv::Size(k, 32), 0.f, 0.f, cv::INTER_LINEAR);
-      cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0, int(100 - k),
-                         cv::BORDER_CONSTANT, {127, 127, 127});
-    }
-  }
+
+  cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
+             cv::INTER_LINEAR);
+  cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0,
+                     int(imgW - resize_img.cols), cv::BORDER_CONSTANT,
+                     {127, 127, 127});
 }
 
 void ClsResizeImg::Run(const cv::Mat &img, cv::Mat &resize_img,
@@ -142,15 +125,11 @@ void ClsResizeImg::Run(const cv::Mat &img, cv::Mat &resize_img,
   else
     resize_w = int(ceilf(imgH * ratio));
 
-  if (!use_tensorrt) {
-    cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
-               cv::INTER_LINEAR);
-    if (resize_w < imgW) {
-      cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0, imgW - resize_w,
-                         cv::BORDER_CONSTANT, cv::Scalar(0, 0, 0));
-    }
-  } else {
-    cv::resize(img, resize_img, cv::Size(100, 32), 0.f, 0.f, cv::INTER_LINEAR);
+  cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
+             cv::INTER_LINEAR);
+  if (resize_w < imgW) {
+    cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0, imgW - resize_w,
+                       cv::BORDER_CONSTANT, cv::Scalar(0, 0, 0));
   }
 }
 

deploy/cpp_infer/src/utility.cpp

@@ -12,12 +12,14 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#include <dirent.h>
+#include <include/utility.h>
 #include <iostream>
 #include <ostream>
+#include <sys/stat.h>
+#include <sys/types.h>
 #include <vector>
 
-#include <include/utility.h>
-
 namespace PaddleOCR {
 
 std::vector<std::string> Utility::ReadDict(const std::string &path) {
@@ -57,4 +59,37 @@ void Utility::VisualizeBboxes(
             << std::endl;
 }
 
+// list all files under a directory
+void Utility::GetAllFiles(const char *dir_name,
+                          std::vector<std::string> &all_inputs) {
+  if (NULL == dir_name) {
+    std::cout << " dir_name is null ! " << std::endl;
+    return;
+  }
+  struct stat s;
+  lstat(dir_name, &s);
+  if (!S_ISDIR(s.st_mode)) {
+    std::cout << "dir_name is not a valid directory !" << std::endl;
+    all_inputs.push_back(dir_name);
+    return;
+  } else {
+    struct dirent *filename; // return value for readdir()
+    DIR *dir;                // return value for opendir()
+    dir = opendir(dir_name);
+    if (NULL == dir) {
+      std::cout << "Can not open dir " << dir_name << std::endl;
+      return;
+    }
+    std::cout << "Successfully opened the dir !" << std::endl;
+    while ((filename = readdir(dir)) != NULL) {
+      if (strcmp(filename->d_name, ".") == 0 ||
+          strcmp(filename->d_name, "..") == 0)
+        continue;
+      // img_dir + std::string("/") + all_inputs[0];
+      all_inputs.push_back(dir_name + std::string("/") +
+                           std::string(filename->d_name));
+    }
+  }
+}
+
 } // namespace PaddleOCR
\ No newline at end of file

deploy/cpp_infer/tools/build.sh

@@ -12,9 +12,10 @@ cmake .. \
     -DWITH_MKL=ON \
     -DWITH_GPU=OFF \
     -DWITH_STATIC_LIB=OFF \
-    -DUSE_TENSORRT=OFF \
+    -DWITH_TENSORRT=OFF \
     -DOPENCV_DIR=${OPENCV_DIR} \
     -DCUDNN_LIB=${CUDNN_LIB_DIR} \
     -DCUDA_LIB=${CUDA_LIB_DIR} \
+    -DTENSORRT_DIR=${TENSORRT_DIR} \
 
 make -j

deploy/cpp_infer/tools/config.txt

@@ -20,10 +20,10 @@ cls_thresh  0.9
 
 # rec config
 rec_model_dir  ./inference/ch_ppocr_mobile_v2.0_rec_infer/
-char_list_file ../../ppocr/utils/ppocr_keys_v1.txt
+char_list_file  ../../ppocr/utils/ppocr_keys_v1.txt
 
 # show the detection results
-visualize 1
+visualize 0
 
 # use_tensorrt
 use_tensorrt 0

deploy/lite/readme_en.md

@@ -6,7 +6,7 @@ paddle-lite is a lightweight inference engine for PaddlePaddle. It provides effi
 
 ## 1. Preparation
 
-### 运行准备
+### Preparation environment
 
 - Computer (for Compiling Paddle Lite)
 - Mobile phone (arm7 or arm8)
@@ -87,8 +87,8 @@ The following table also provides a series of models that can be deployed on mob
 
 |Version|Introduction|Model size|Detection model|Text Direction model|Recognition model|Paddle-Lite branch|
 |---|---|---|---|---|---|---|
-|V2.0|extra-lightweight chinese OCR optimized model|7.8M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[download lin](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[download lin](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
-|V2.0(slim)|extra-lightweight chinese OCR optimized model|3.3M|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
+|V2.0|extra-lightweight chinese OCR optimized model|7.8M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
+|V2.0(slim)|extra-lightweight chinese OCR optimized model|3.3M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
 
 If you directly use the model in the above table for deployment, you can skip the following steps and directly read [Section 2.2](#2.2 Run optimized model on Phone).
 

doc/doc_en/models_list_en.md

@@ -103,14 +103,14 @@ python3 generate_multi_language_configs.py -l it \
 | german_mobile_v2.0_rec | ppocr/utils/dict/german_dict.txt | Lightweight model for German recognition|[rec_german_lite_train.yml](../../configs/rec/multi_language/rec_german_lite_train.yml)|2.65M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/german_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/german_mobile_v2.0_rec_train.tar) |
 | korean_mobile_v2.0_rec | ppocr/utils/dict/korean_dict.txt | Lightweight model for Korean recognition|[rec_korean_lite_train.yml](../../configs/rec/multi_language/rec_korean_lite_train.yml)|3.9M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_train.tar) |
 | japan_mobile_v2.0_rec | ppocr/utils/dict/japan_dict.txt | Lightweight model for Japanese recognition|[rec_japan_lite_train.yml](../../configs/rec/multi_language/rec_japan_lite_train.yml)|4.23M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/japan_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/japan_mobile_v2.0_rec_train.tar) |
-| chinese_cht_mobile_v2.0_rec | ppocr/utils/dict/chinese_cht_dict.txt | Lightweight model for chinese cht recognition|rec_chinese_cht_lite_train.yml|5.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_train.tar) |
+| chinese_cht_mobile_v2.0_rec | ppocr/utils/dict/chinese_cht_dict.txt | Lightweight model for chinese cht recognition|rec_chinese_cht_lite_train.yml|5.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_train.tar) |
 | te_mobile_v2.0_rec | ppocr/utils/dict/te_dict.txt | Lightweight model for Telugu recognition|rec_te_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_train.tar) |
 | ka_mobile_v2.0_rec | ppocr/utils/dict/ka_dict.txt | Lightweight model for Kannada recognition|rec_ka_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_train.tar) |
 | ta_mobile_v2.0_rec | ppocr/utils/dict/ta_dict.txt | Lightweight model for Tamil recognition|rec_ta_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_train.tar) |
-| latin_mobile_v2.0_rec | ppocr/utils/dict/latin_dict.txt | Lightweight model for latin recognition |  [rec_latin_lite_train.yml](../../configs/rec/multi_language/rec_latin_lite_train.yml) |2.6M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/latin_ppocr_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/latin_ppocr_mobile_v2.0_rec_train.tar) |
-| arabic_mobile_v2.0_rec |  ppocr/utils/dict/arabic_dict.txt | Lightweight model for arabic recognition | [rec_arabic_lite_train.yml](../../configs/rec/multi_language/rec_arabic_lite_train.yml) |2.6M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/arabic_ppocr_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/arabic_ppocr_mobile_v2.0_rec_train.tar) |
-| cyrillic_mobile_v2.0_rec | ppocr/utils/dict/cyrillic_dict.txt | Lightweight model for cyrillic recognition | [rec_cyrillic_lite_train.yml](../../configs/rec/multi_language/rec_cyrillic_lite_train.yml) |2.6M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/cyrillic_ppocr_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/cyrillic_ppocr_mobile_v2.0_rec_train.tar) |
-| devanagari_mobile_v2.0_rec | ppocr/utils/dict/devanagari_dict.txt | Lightweight model for devanagari recognition | [rec_devanagari_lite_train.yml](../../configs/rec/multi_language/rec_devanagari_lite_train.yml) |2.6M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/devanagari_ppocr_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/devanagari_ppocr_mobile_v2.0_rec_train.tar) |
+| latin_mobile_v2.0_rec | ppocr/utils/dict/latin_dict.txt | Lightweight model for latin recognition |  [rec_latin_lite_train.yml](../../configs/rec/multi_language/rec_latin_lite_train.yml) |2.6M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/latin_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/latin_ppocr_mobile_v2.0_rec_train.tar) |
+| arabic_mobile_v2.0_rec |  ppocr/utils/dict/arabic_dict.txt | Lightweight model for arabic recognition | [rec_arabic_lite_train.yml](../../configs/rec/multi_language/rec_arabic_lite_train.yml) |2.6M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/arabic_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/arabic_ppocr_mobile_v2.0_rec_train.tar) |
+| cyrillic_mobile_v2.0_rec | ppocr/utils/dict/cyrillic_dict.txt | Lightweight model for cyrillic recognition | [rec_cyrillic_lite_train.yml](../../configs/rec/multi_language/rec_cyrillic_lite_train.yml) |2.6M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/cyrillic_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/cyrillic_ppocr_mobile_v2.0_rec_train.tar) |
+| devanagari_mobile_v2.0_rec | ppocr/utils/dict/devanagari_dict.txt | Lightweight model for devanagari recognition | [rec_devanagari_lite_train.yml](../../configs/rec/multi_language/rec_devanagari_lite_train.yml) |2.6M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/devanagari_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/devanagari_ppocr_mobile_v2.0_rec_train.tar) |
 
 For more supported languages, please refer to : [Multi-language model](./multi_languages_en.md)
 

tools/infer/utility.py

@@ -21,6 +21,9 @@ import json
 from PIL import Image, ImageDraw, ImageFont
 import math
 from paddle import inference
+import time
+from ppocr.utils.logging import get_logger
+logger = get_logger()
 
 
 def str2bool(v):
@@ -40,6 +43,7 @@ inference_args_list = [
     ['total_process_num', int, 1],
     ['process_id', int, 0],
     ['gpu_mem', int, 500],
+    ['cpu_threads', int, 10],
     # params for text detector
     ['image_dir', str, None],
     ['det_algorithm', str, 'DB'],
@@ -128,19 +132,97 @@ def create_predictor(args, mode, logger):
         config.enable_use_gpu(args.gpu_mem, 0)
         if args.use_tensorrt:
             config.enable_tensorrt_engine(
-                precision_mode=inference.PrecisionType.Half
-                if args.use_fp16 else inference.PrecisionType.Float32,
-                max_batch_size=args.max_batch_size)
+                precision_mode=inference.PrecisionType.Float32,
+                max_batch_size=args.max_batch_size,
+                min_subgraph_size=3)  # skip the minmum trt subgraph
+        if mode == "det" and "mobile" in model_file_path:
+            min_input_shape = {
+                "x": [1, 3, 50, 50],
+                "conv2d_92.tmp_0": [1, 96, 20, 20],
+                "conv2d_91.tmp_0": [1, 96, 10, 10],
+                "nearest_interp_v2_1.tmp_0": [1, 96, 10, 10],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 20, 20],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 20, 20],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 20, 20],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 20, 20],
+                "elementwise_add_7": [1, 56, 2, 2],
+                "nearest_interp_v2_0.tmp_0": [1, 96, 2, 2]
+            }
+            max_input_shape = {
+                "x": [1, 3, 2000, 2000],
+                "conv2d_92.tmp_0": [1, 96, 400, 400],
+                "conv2d_91.tmp_0": [1, 96, 200, 200],
+                "nearest_interp_v2_1.tmp_0": [1, 96, 200, 200],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 400, 400],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 400, 400],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 400, 400],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 400, 400],
+                "elementwise_add_7": [1, 56, 400, 400],
+                "nearest_interp_v2_0.tmp_0": [1, 96, 400, 400]
+            }
+            opt_input_shape = {
+                "x": [1, 3, 640, 640],
+                "conv2d_92.tmp_0": [1, 96, 160, 160],
+                "conv2d_91.tmp_0": [1, 96, 80, 80],
+                "nearest_interp_v2_1.tmp_0": [1, 96, 80, 80],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 160, 160],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 160, 160],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 160, 160],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 160, 160],
+                "elementwise_add_7": [1, 56, 40, 40],
+                "nearest_interp_v2_0.tmp_0": [1, 96, 40, 40]
+            }
+        if mode == "det" and "server" in model_file_path:
+            min_input_shape = {
+                "x": [1, 3, 50, 50],
+                "conv2d_59.tmp_0": [1, 96, 20, 20],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 20, 20],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 20, 20],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 20, 20],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 20, 20]
+            }
+            max_input_shape = {
+                "x": [1, 3, 2000, 2000],
+                "conv2d_59.tmp_0": [1, 96, 400, 400],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 400, 400],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 400, 400],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 400, 400],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 400, 400]
+            }
+            opt_input_shape = {
+                "x": [1, 3, 640, 640],
+                "conv2d_59.tmp_0": [1, 96, 160, 160],
+                "nearest_interp_v2_2.tmp_0": [1, 96, 160, 160],
+                "nearest_interp_v2_3.tmp_0": [1, 24, 160, 160],
+                "nearest_interp_v2_4.tmp_0": [1, 24, 160, 160],
+                "nearest_interp_v2_5.tmp_0": [1, 24, 160, 160]
+            }
+        elif mode == "rec":
+            min_input_shape = {"x": [args.rec_batch_num, 3, 32, 10]}
+            max_input_shape = {"x": [args.rec_batch_num, 3, 32, 2000]}
+            opt_input_shape = {"x": [args.rec_batch_num, 3, 32, 320]}
+        elif mode == "cls":
+            min_input_shape = {"x": [args.rec_batch_num, 3, 48, 10]}
+            max_input_shape = {"x": [args.rec_batch_num, 3, 48, 2000]}
+            opt_input_shape = {"x": [args.rec_batch_num, 3, 48, 320]}
+        else:
+            min_input_shape = {"x": [1, 3, 10, 10]}
+            max_input_shape = {"x": [1, 3, 1000, 1000]}
+            opt_input_shape = {"x": [1, 3, 500, 500]}
+        config.set_trt_dynamic_shape_info(min_input_shape, max_input_shape,
+                                          opt_input_shape)
+
     else:
         config.disable_gpu()
-        config.set_cpu_math_library_num_threads(6)
+        if hasattr(args, "cpu_threads"):
+            config.set_cpu_math_library_num_threads(args.cpu_threads)
+        else:
+            config.set_cpu_math_library_num_threads(
+                10)  # default cpu threads as 10
         if args.enable_mkldnn:
             # cache 10 different shapes for mkldnn to avoid memory leak
             config.set_mkldnn_cache_capacity(10)
             config.enable_mkldnn()
-            #  TODO LDOUBLEV: fix mkldnn bug when bach_size  > 1
-            # config.set_mkldnn_op({'conv2d', 'depthwise_conv2d', 'pool2d', 'batch_norm'})
-            args.rec_batch_num = 1
 
     # enable memory optim
     config.enable_memory_optim()
@@ -203,7 +285,7 @@ def draw_ocr(image,
              txts=None,
              scores=None,
              drop_score=0.5,
-             font_path="./doc/simfang.ttf"):
+             font_path="./doc/fonts/simfang.ttf"):
     """
     Visualize the results of OCR detection and recognition
     args:
@@ -411,22 +493,4 @@ def draw_boxes(image, boxes, scores=None, drop_score=0.5):
 
 
 if __name__ == '__main__':
-    test_img = "./doc/test_v2"
-    predict_txt = "./doc/predict.txt"
-    f = open(predict_txt, 'r')
-    data = f.readlines()
-    img_path, anno = data[0].strip().split('\t')
-    img_name = os.path.basename(img_path)
-    img_path = os.path.join(test_img, img_name)
-    image = Image.open(img_path)
-
-    data = json.loads(anno)
-    boxes, txts, scores = [], [], []
-    for dic in data:
-        boxes.append(dic['points'])
-        txts.append(dic['transcription'])
-        scores.append(round(dic['scores'], 3))
-
-    new_img = draw_ocr(image, boxes, txts, scores)
-
-    cv2.imwrite(img_name, new_img)
+    pass