Linux集成文档-C

cd src
mkdir build && cd build
cmake .. && make
./easyedge_image_inference {模型RES文件夹}  {测试图片路径}
# 如果是NNIE引擎，使用sudo运行
sudo ./easyedge_image_inference {模型RES文件夹}  {测试图片路径}

cmake -DEDGE_BUILD_OPENCV=ON

 > ./easyedge_image_inference ../../../../RES 2.jpeg
2019-02-13 16:46:12,659 INFO [EasyEdge] [easyedge.cpp:34] 140606189016192 Baidu EasyEdge Linux Development Kit 0.2.1(20190213)
2019-02-13 16:46:14,083 INFO [EasyEdge] [paddlev2_edge_predictor.cpp:60] 140606189016192 Allocate graph success.
2019-02-13 16:46:14,326 DEBUG [EasyEdge] [paddlev2_edge_predictor.cpp:143] 140606189016192 Inference costs 168 ms
1, 1:txt_frame, p:0.994905 loc: 0.168161, 0.153654, 0.920856, 0.779621
Done

# ./easyedge_serving {res_dir} {serial_key} {host, default 0.0.0.0} {port, default 24401}
 ./easyedge_serving ../../../RES "1111-1111-1111-1111" 0.0.0.0  24401

HTTP is now serving at 0.0.0.0:24401

    // step 1: 配置模型资源目录
    EdgePredictorConfig config;
    config.model_dir = {模型文件目录};
    
    // step 2: 创建并初始化Predictor；在这里选择合适的引擎
    auto predictor = global_controller()->CreateEdgePredictor(config);


    // step 3-1: 预测图像
    auto img = cv::imread({图片路径});
    std::vector<EdgeResultData> results;
    predictor->infer(img, results);
	
	// step 3-2: 预测视频
	std::vector<EdgeResultData> results;
	FrameTensor frame_tensor;
    VideoConfig video_config;
    video_config.source_type = static_cast<SourceType>(video_type);  // source_type 定义参考头文件 easyedge_video.h
    video_config.source_value = video_src;
    /*
    ... more video_configs, 根据需要配置video_config的各选项
    */
    auto video_decoding = CreateVideoDecoding(video_config);
    while (video_decoding->next(frame_tensor) == EDGE_OK) {
        results.clear();
        if (frame_tensor.is_needed) {
            predictor->infer(frame_tensor.frame, results);
            render(frame_tensor.frame, results, predictor->model_info().kind);
        }
        //video_decoding->display(frame_tensor); // 显示当前frame，需在video_config中开启配置
        //video_decoding->save(frame_tensor); // 存储当前frame到视频，需在video_config中开启配置
    }

/**
 * @brief 序列号设置；序列号不设置留空时，SDK将会自动尝试使用本地已经激活成功的有效期内的序列号
 * 值类型：string
 * 默认值：空
 */
static constexpr auto PREDICTOR_KEY_SERIAL_NUM = "PREDICTOR_KEY_SERIAL_NUM";

EdgePredictorConfig config;
config.model_dir = ...;
config.set_config(params::PREDICTOR_KEY_SERIAL_NUM, "1DB7-1111-1111-D27D");

auto predictor = global_controller()->CreateEdgePredictor(config);
predictor->init();

 /**
  * @brief
  * 通用接口
  * @param image: must be BGR , HWC format (opencv default)
  * @param result
  * @return
  */
 virtual int infer(
         cv::Mat& image, std::vector<EdgeResultData>& result
 ) = 0;

 /**
  * @brief
  * 批量图片推理接口
  * @param image: must be BGR , HWC format (opencv default)
  * @param result
  * @return
  */
 virtual int infer(
         std::vector<cv::Mat>& image, std::vector<std::vector<EdgeResultData>>& result
 ) = 0;

struct EdgeResultData {
    int index;  // 分类结果的index
    std::string label;  // 分类结果的label
    float prob;  // 置信度

    // 物体检测、图像分割时才有意义
    float x1, y1, x2, y2;  // (x1, y1): 左上角， （x2, y2): 右下角； 均为0~1的长宽比例值。

    // 图像分割的模型，该字段才有意义
    // 请注意：图像分割时，以下两个字段会比较大，使用完成之后请及时释放EdgeResultData
    cv::Mat mask;  // 0, 1 的mask
    std::string mask_rle;  // Run Length Encoding，游程编码的mask

    // 目标追踪模型，该字段才有意义
    int trackid;  // 轨迹id
    int frame;  // 处于视频中的第几帧
    EdgeTrackStat track_stat;  // 跟踪状态
};

cv::Mat mask为图像掩码的二维数组
{
  {0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
  {0, 0, 0, 1, 1, 1, 0, 0, 0, 0},
  {0, 0, 0, 1, 1, 1, 0, 0, 0, 0},
  {0, 0, 0, 1, 1, 1, 0, 0, 0, 0},
  {0, 0, 0, 1, 1, 1, 0, 0, 0, 0},
  {0, 0, 0, 0, 0, 0, 0, 0, 0, 0},
}
其中1代表为目标区域，0代表非目标区域

EdgeLogConfig log_config;
log_config.enable_debug = true;
global_controller()->set_log_config(log_config);

 /**
     * @brief 开启一个简单的demo http服务。
     * 该方法会block直到收到sigint/sigterm。
     * http服务里，图片的解码运行在cpu之上，可能会降低推理速度。
     * @tparam ConfigT
     * @param config
     * @param host
     * @param port
     * @param service_id service_id  user parameter, uri '/get/service_id' will respond this value with 'text/plain'
     * @param instance_num 实例数量，根据内存/显存/时延要求调整
     * @return
     */
    template<typename ConfigT>
    int start_http_server(
            const ConfigT &config,
            const std::string &host,
            int port,
            const std::string &service_id,
            int instance_num = 1);

import requests

with open('./1.jpg', 'rb') as f:
    img_data = f.read()
    result = requests.post(
	    'http://127.0.0.1:24401/',
	    params={'threshold': 0.1},
	    data=img_data).json()

{
	"image": "<base64数据>"
	"top_num": 5
}

{
	"image": "<base64数据>"
}

{
    "cost_ms": 52,
    "error_code": 0,
    "results": [
        {
            "confidence": 0.94482421875,
            "index": 1,
            "label": "IronMan",
            "x1": 0.059185408055782318,
            "x2": 0.18795496225357056,
            "y1": 0.14762254059314728,
            "y2": 0.52510076761245728
        },
        {
            "confidence": 0.94091796875,
            "index": 1,
            "label": "IronMan",
            "x1": 0.79151463508605957,
            "x2": 0.92310667037963867,
            "y1": 0.045728668570518494,
            "y2": 0.42920106649398804
        }
      ]
}

global_controller()->set_config(easyedge::params::CONTROLLER_KEY_LOG_BRAND, "MY_BRAND");

EdgePredictorConfig config;
config.set_config(easyedge::params::PREDICTOR_KEY_CPU_THREADS_NUM, 4);

EdgePredictorConfig config;
config.set_config(params::PREDICTOR_KEY_VINO_MAX_BATCH_SIZE, 4); //CPU加速版特殊设置

int batch_size = 2; // 使用前修改batch_size再编译、执行
while (get_next_batch(imgs, img_files, batch_size, start_index)) {
    ...
}

find_package(CURL REQUIRED)
target_link_libraries(easyedge_demo  ${OpenCV_LIBS} easyedge_static pthread ${CURL_LIBRARIES} verify_static ${其他需要的库})

paddle.fluid.core.EnforceNotMet: Enforce failed. Expected allocating <= available, but received allocating:20998686233 > available:19587333888.
Insufficient GPU memory to allocation. at [/paddle/paddle/fluid/platform/gpu_info.cc:170]

open sys: Permission denied
open err
: Permission denied
open err
: Permission denied

export HTTP_PROXY="http://192.168.1.100:8888"
./easyedge_demo ...

headers = curl_slist_append(headers, "Expect:");

LD_LIBRARY_PATH=$LD_LIBRARY_PATH:../lib ./easyedge_demo

export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:../thirdparty/opencv/lib
(tips: 上面冒号后面接的thirdparty/opencv/lib路径以实际项目中路径为准，比如也可能是../../thirdparty/opencv/lib)

--   Video I/O:
--     DC1394:                      YES (ver 2.2.4)
--     FFMPEG:                      YES
--       avcodec:                   YES (ver 56.60.100)
--       avformat:                  YES (ver 56.40.101)
--       avutil:                    YES (ver 54.31.100)
--       swscale:                   YES (ver 3.1.101)
--       avresample:                NO
--     libv4l/libv4l2:              NO
--     v4l/v4l2:                    linux/videodev2.h

apt-get install yasm libjpeg-dev libjasper-dev libavcodec-dev libavformat-dev libswscale-dev libdc1394-22-dev libgstreamer0.10-dev libgstreamer-plugins-base0.10-dev libv4l-dev python-dev python-numpy libtbb-dev libqt4-dev libgtk2.0-dev libfaac-dev libmp3lame-dev libopencore-amrnb-dev libopencore-amrwb-dev libtheora-dev libvorbis-dev libxvidcore-dev x264 v4l-utils ffmpeg