C++ 利用Paddle进行文字识别

1、OCR Paddle开源地址：PaddlePaddle · GitHub

2、OpenCV开源地址：OpenCV · GitHub

3、示例：

#include <iostream>
#include <vector>
#include <opencv2/opencv.hpp>
#include <paddle_api.h>

int main() {
    // 读取图像
    cv::Mat image = cv::imread("test.jpg");

    // 初始化Paddle Inference引擎
    std::string model_dir = "your/model/directory";
    std::string pdmodel_file = model_dir + "/model.pdmodel";
    std::string pdiparams_file = model_dir + "/model.pdiparams";
    paddle::lite_api::MobileConfig config;
    config.set_model_from_file(pdmodel_file, pdiparams_file);
    config.set_threads(1);  // 设置线程数
    config.set_power_mode(paddle::lite_api::LITE_POWER_HIGH);  // 设置运行模式
    auto predictor = paddle::lite_api::CreatePaddlePredictor(config);

    // 图像预处理
    cv::cvtColor(image, image, cv::COLOR_BGR2RGB);
    cv::resize(image, image, cv::Size(320, 320));
    
    // 将图像转换为PaddleTensor输入格式
    paddle::lite_api::Tensor input_tensor;
    input_tensor.Resize({1, 3, image.rows, image.cols});
    input_tensor.SetData(reinterpret_cast<const float*>(image.data), image.total() * image.elemSize());

    // 进行文字识别
    paddle::lite_api::Tensor output_tensor;
    predictor->Run({input_tensor}, &output_tensor);

    // 解析识别结果
    std::vector<int64_t> output_shape = output_tensor.shape();
    int64_t output_size = std::accumulate(output_shape.begin(), output_shape.end(), 1, std::multiplies<int64_t>());
    const float* output_data = output_tensor.data<float>();
    
    // 输出识别结果
    for (int i = 0; i < output_size; i++) {
        std::cout << "识别结果: " << output_data[i] << std::endl;
    }

    return 0;
}