EasyEdge平台Atlas-300适配SDK开发文档

sudo apt-get install sshpass build-essential
升级安装CANN5.0.3，涉及到对应的固件及驱动更新，按照昇腾https://www.hiascend.com官方指导安装验证

cd src
mkdir build && cd build
cmake .. && make
sudo ldconfig
./demo_image_inference/easyedge_image_inference {模型RES文件夹}  {测试图片路径或仅包含图片的文件夹路径} {序列号}

export LD_LIBRARY_PATH=path_to_sdk/lib/:path_to_sdk/thirdparty/opencv/lib/:$LD_LIBRARY_PATH
./demo_image_inference/easyedge_image_inference ../../../../RES/ /home/images/mix008.jpeg
2020-08-06 20:56:30,665 INFO [EasyEdge] 548125646864 Compiling model for fast inference, this may take a while
...
Results of image /ljay/images/mix008.jpeg: 
2, kiwi, p:0.997594 loc: 0.352087, 0.56119, 0.625748, 0.868399
2, kiwi, p:0.993221 loc: 0.45789, 0.0730294, 0.73641, 0.399429
2, kiwi, p:0.992884 loc: 0.156876, 0.0598725, 0.3802, 0.394706
1, tomato, p:0.992125 loc: 0.523592, 0.389156, 0.657738, 0.548069
1, tomato, p:0.991821 loc: 0.665461, 0.419503, 0.805282, 0.573558
1, tomato, p:0.989883 loc: 0.297427, 0.439999, 0.432197, 0.59325
1, tomato, p:0.981654 loc: 0.383444, 0.248203, 0.506606, 0.400926
1, tomato, p:0.971682 loc: 0.183775, 0.556587, 0.286996, 0.711361
1, tomato, p:0.968722 loc: 0.379391, 0.0386965, 0.51672, 0.209681
Done

export LD_LIBRARY_PATH=path_to_sdk/lib/:path_to_sdk/thirdparty/opencv/lib/:$LD_LIBRARY_PATH
# ./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 0: 设置序列号
    global_controller()->set_licence_key("set your license here");

    // step 1: 配置模型资源目录
    AtlasConfig config; 
    config.model_dir = {模型文件目录};

    // step 2: 创建并初始化Predictor； 
    auto predictor = global_controller()->CreateEdgePredictor(config);
    int ret = predictor->init();
    # 若返回非0，请查看输出日志排查错误原因。
    auto img = cv::imread({图片路径});
    // step 3: 预测图像
    std::vector<EdgeResultData> result2;
    predictor->infer(img, result2);
    # 解析result2即可获取结果

// step 0: 设置序列号
global_controller()->set_licence_key("set your license here");

// step 1: 配置模型资源目录
AtlasConfig config; 
config.model_dir = {模型文件目录};

// step 3: 创建Predictor；这这里选择合适的引擎
auto predictor = global_controller()->CreateEdgePredictor(config);

// step 4: 设置异步回调
predictor->set_result_handler(YOUR_HANDLER);

// step 5: 初始化
int ret = predictor->init();
# 若返回非0，请查看输出日志排查错误原因。

// step 6: 预测图像
auto img = cv::imread({图片路径});
color_format = kBGR;
float threshold = 0.1;

uint64_t seq_id;
predictor->infer_async(img, color_format, 0.1, nullptr, seq_id);
# YOUR_HANDLER里面有seq_id的回调结果

virtual int set_licence_key(const std::string& license) = 0;

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