在agx orin部署测试paddleocr(c++版本)

paddleocr的原版CMakeLists编译涉及到要配置的环境变量太多，里面都是用变量代替，牵一发而动全身，十分繁琐，于是我重写CMakeLists,只用了简洁而必要的命令，很快便能成功运行整个项目。

cmake_minimum_required(VERSION 3.5)
# 设置c++标准
set(CMAKE_CXX_STANDARD 17)
project(PaddleOcr)
# 默认arm
set(path lib/arm)
set(opencvVersion opencv410)#设置opencv版本
# 头文件
include_directories(/home/nvidia/paddleOCR/PaddleOCR-release-2.6/deploy/cpp_infer/include)
include_directories(/home/nvidia/paddleOCR/PaddleOCR-release-2.6/deploy/cpp_infer)
include_directories(/usr/include)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/paddle/include)
include_directories(/home/nvidia/opencv-4.1.0/include/opencv2)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/glog/include)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/AutoLog-main)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/gflags/include)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/protobuf/include)
include_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/threadpool)
# 库文件
link_directories(/usr/lib)
link_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/glog/lib)
link_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/paddle/lib)
link_directories(/home/nvidia/opencv-4.1.0/build/lib)
link_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/protobuf/lib)
link_directories(/home/nvidia/paddleOCR/paddle_inference_install_dir/third_party/install/gflags/lib)

aux_source_directory (src SRC_LIST)
add_executable (main ${SRC_LIST})
# c++17
target_link_libraries(main  opencv_highgui opencv_core opencv_imgproc opencv_imgcodecs opencv_calib3d opencv_features2d opencv_videoio protobuf glog gflags paddle_inference pthread)
# 注意测试
set (EXECUTABLE_OUTPUT_PATH ${PROJECT_SOURCE_DIR}/bin/arm)

• 更改图片输出地址
  main.cpp 115行
  

• 改变识别对应的字典
  在args.cpp49 行
  

• 在detection框上添加预测出来的文字
  utility.cpp 61行
  

识别命令

检测
./main --det_model_dir=/home/nvidia/paddleOCR/inference/detection/en_PP-OCRv3_det_infer --image_dir=/home/nvidia/paddleOCR/imgae/En_Num/0.jpg     --det=true   --rec=false

检测+识别
./main  --det_model_dir=/home/nvidia/paddleOCR/inference/detection/en_PP-OCRv3_det_infer  --rec_model_dir=/home/nvidia/paddleOCR/inference/recognize/en_PP-OCRv3_rec_infer   --image_dir=/home/nvidia/paddleOCR/imgae/En_Num/0.jpg     --use_angle_cls=false    --det=true     --rec=true    --cls=false 

中文检测+中文识别
./main  --det_model_dir=/home/nvidia/paddleOCR/inference/detection/ch_ppocr_server_v2.0_det_infer  --rec_model_dir=/home/nvidia/paddleOCR/inference/recognize/ch_ppocr_server_v2.0_rec_infer   --image_dir=/home/nvidia/paddleOCR/imgae/ch     --use_angle_cls=false    --det=true     --rec=true    --cls=false 

检测+识别+方向分类器
./main  --det_model_dir=/home/nvidia/paddleOCR/inference/detection/en_PP-OCRv3_det_infer  --rec_model_dir=/home/nvidia/paddleOCR/inference/recognize/en_PP-OCRv3_rec_infer      --cls_model_dir=/home/nvidia/paddleOCR/inference/cls/ch_ppocr_mobile_v2.0_cls_infer  --image_dir=/home/nvidia/paddleOCR/imgae/En_Num/0.jpg     --use_angle_cls=true    --det=true     --rec=true    --cls=true

ocr_rec.cpp

#include 

namespace PaddleOCR {

void CRNNRecognizer::Run(std::vector<cv::Mat> img_list,
                         std::vector<std::string> &rec_texts,
                         std::vector<float> &rec_text_scores,
                         std::vector<double> &times) {
  std::chrono::duration<float> preprocess_diff =
      std::chrono::steady_clock::now() - std::chrono::steady_clock::now();
  std::chrono::duration<float> inference_diff =
      std::chrono::steady_clock::now() - std::chrono::steady_clock::now();
  std::chrono::duration<float> postprocess_diff =
      std::chrono::steady_clock::now() - std::chrono::steady_clock::now();

  int img_num = img_list.size();
  std::vector<float> width_list;
  for (int i = 0; i < img_num; i++) {
    width_list.push_back(float(img_list[i].cols) / img_list[i].rows);
  }
  std::vector<int> indices = Utility::argsort(width_list);

  for (int beg_img_no = 0; beg_img_no < img_num;
       beg_img_no += this->rec_batch_num_) {
    auto preprocess_start = std::chrono::steady_clock::now();
    int end_img_no = std::min(img_num, beg_img_no + this->rec_batch_num_);
    int batch_num = end_img_no - beg_img_no;
    int imgH = this->rec_image_shape_[1];
    int imgW = this->rec_image_shape_[2];
    float max_wh_ratio = imgW * 1.0 / imgH;
    for (int ino = beg_img_no; ino < end_img_no; ino++) {
      int h = img_list[indices[ino]].rows;
      int w = img_list[indices[ino]].cols;
      float wh_ratio = w * 1.0 / h;
      max_wh_ratio = std::max(max_wh_ratio, wh_ratio);
    }

    int batch_width = imgW;
    std::vector<cv::Mat> norm_img_batch;
    for (int ino = beg_img_no; ino < end_img_no; ino++) {
      cv::Mat srcimg;
      img_list[indices[ino]].copyTo(srcimg);
      cv::Mat resize_img;
      this->resize_op_.Run(srcimg, resize_img, max_wh_ratio,
                           this->use_tensorrt_, this->rec_image_shape_);
      this->normalize_op_.Run(&resize_img, this->mean_, this->scale_,
                              this->is_scale_);
      norm_img_batch.push_back(resize_img);
      batch_width = std::max(resize_img.cols, batch_width);
    }

    std::vector<float> input(batch_num * 3 * imgH * batch_width, 0.0f);
    this->permute_op_.Run(norm_img_batch, input.data());
    auto preprocess_end = std::chrono::steady_clock::now();
    preprocess_diff += preprocess_end - preprocess_start;
    // Inference.
    auto input_names = this->predictor_->GetInputNames();
    auto input_t = this->predictor_->GetInputHandle(input_names[0]);
    input_t->Reshape({batch_num, 3, imgH, batch_width});
    auto inference_start = std::chrono::steady_clock::now();
    input_t->CopyFromCpu(input.data());
    this->predictor_->Run();

    std::vector<float> predict_batch;
    auto output_names = this->predictor_->GetOutputNames();
    auto output_t = this->predictor_->GetOutputHandle(output_names[0]);
    auto predict_shape = output_t->shape();

    int out_num = std::accumulate(predict_shape.begin(), predict_shape.end(), 1,
                                  std::multiplies<int>());
    predict_batch.resize(out_num);
    // predict_batch is the result of Last FC with softmax
    output_t->CopyToCpu(predict_batch.data());
    auto inference_end = std::chrono::steady_clock::now();
    inference_diff += inference_end - inference_start;
    // ctc decode
    auto postprocess_start = std::chrono::steady_clock::now();
    for (int m = 0; m < predict_shape[0]; m++) {
      std::string str_res;
      int argmax_idx;
      int last_index = 0;
      float score = 0.f;
      int count = 0;
      float max_value = 0.0f;


      for (int n = 0; n < predict_shape[1]; n++) {
        // get idx
        argmax_idx = int(Utility::argmax(
            &predict_batch[(m * predict_shape[1] + n) * predict_shape[2]],
            &predict_batch[(m * predict_shape[1] + n + 1) * predict_shape[2]]));
        // get score
        max_value = float(*std::max_element(
            &predict_batch[(m * predict_shape[1] + n) * predict_shape[2]],
            &predict_batch[(m * predict_shape[1] + n + 1) * predict_shape[2]]));

/*针对en_dict.txt这个字典，0~9数字，10~16,43~48,75~93符号,17~42大写字母,49~74小写字母，*/
          /*字典打印*/
          // for(int mm=0;mm<100;mm++)
          // {
          //   std::cout<<"数字： "<
          // }
        /*只识别数字*/
        //if(argmax_idx<11  || (argmax_idx==96))
        /*只识别小写字母*/
         //if(argmax_idx==0 || (argmax_idx<76 && argmax_idx>49)  || (argmax_idx==96))
        /*只识别大写字母*/
        //if(argmax_idx==0 ||  ((argmax_idx>17) && (argmax_idx<44))  || (argmax_idx==96))
        /*识别大写字母和数字*/
        // if(argmax_idx==0 ||  ((argmax_idx>17) && (argmax_idx<44)) || (argmax_idx<11) || (argmax_idx==96))
        // {
          // std::cout<<"--------------------1---------------"<
          // std::cout<<" argmax_idx "<
          if (argmax_idx > 0 && (!(n > 0 && argmax_idx == last_index)))
          {
            // std::cout<<"--------------------2---------------"<
            // std::cout<<" argmax_idx "<
            score += max_value;
            count += 1;
            str_res += label_list_[argmax_idx];
          }
        // }
        // else
        // {
        //    continue; 
        // }
         last_index = argmax_idx;
      }
      score /= count;
      if (std::isnan(score)) {
        continue;
      }
      rec_texts[indices[beg_img_no + m]] = str_res;
      rec_text_scores[indices[beg_img_no + m]] = score;
    }
    auto postprocess_end = std::chrono::steady_clock::now();
    postprocess_diff += postprocess_end - postprocess_start;
  }
  times.push_back(double(preprocess_diff.count() * 1000));
  times.push_back(double(inference_diff.count() * 1000));
   times.push_back(double(postprocess_diff.count() * 1000));
}

void CRNNRecognizer::LoadModel(const std::string &model_dir) {
  paddle_infer::Config config;
  config.SetModel(model_dir + "/inference.pdmodel",
                  model_dir + "/inference.pdiparams");
  std::cout << "In PP-OCRv3, default rec_img_h is 48,"
            << "if you use other model, you should set the param rec_img_h=32"
            << std::endl;
  if (this->use_gpu_) {
    config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
    if (this->use_tensorrt_) {
      auto precision = paddle_infer::Config::Precision::kFloat32;
      if (this->precision_ == "fp16") {
        precision = paddle_infer::Config::Precision::kHalf;
      }
      if (this->precision_ == "int8") {
        precision = paddle_infer::Config::Precision::kInt8;
      }
      if (!Utility::PathExists("./trt_rec_shape.txt")) {
        config.CollectShapeRangeInfo("./trt_rec_shape.txt");
      } else {
        config.EnableTunedTensorRtDynamicShape("./trt_rec_shape.txt", true);
      }
    }
  } else {
    config.DisableGpu();
    if (this->use_mkldnn_) {
      config.EnableMKLDNN();
      // cache 10 different shapes for mkldnn to avoid memory leak
      config.SetMkldnnCacheCapacity(10);
    }
    config.SetCpuMathLibraryNumThreads(this->cpu_math_library_num_threads_);
  }

  // get pass_builder object
  auto pass_builder = config.pass_builder();
  // delete "matmul_transpose_reshape_fuse_pass"
  pass_builder->DeletePass("matmul_transpose_reshape_fuse_pass");
  config.SwitchUseFeedFetchOps(false);
  // true for multiple input
  config.SwitchSpecifyInputNames(true);

  config.SwitchIrOptim(true);

  config.EnableMemoryOptim();
  //   config.DisableGlogInfo();

  this->predictor_ = paddle_infer::CreatePredictor(config);
}

} // namespace PaddleOCR