视觉任务服务器端LinuxSDK集成文档-Python

pip3 install -U paddlepaddle

pip3 install -U paddlepaddle-gpu

source /opt/intel/openvino/bin/setupvars.sh

pip3 install -U BaiduAI_EasyEdge_SDK-{版本号}-cp36-cp36m-linux_x86_64.whl

pred = edge.Program()
pred.set_auth_license_key("这里填写序列号")

cd ${SDK_ROOT}

# 1. 安装 python wheel 包
tar -xzvf python/*.tar.gz
pip install -U {对应 Python 版本的 wheel 包}

# 2. 设置 LD_LIBRARY_PATH
tar -xzvf cpp/*.tar.gz 
export EDGE_ROOT=$(readlink -f $(ls -h | grep "baidu_easyedge_linux_cpp"))
export LD_LIBRARY_PATH=$EDGE_ROOT/lib

# 3. 运行 demo
python3 demo.py {RES文件夹路径}  {测试图片路径}

ImportError: libeasyedge.so.0.4.3: cannot open shared object file: No such file or directory
ImportError: libopencv_core.so.3.4: cannot open shared object file: No such file or directory

python3 demo.py {model_dir} {image_name.jpg}

import BaiduAI.EasyEdge as edge

pred = edge.Program()
pred.set_auth_license_key("这里填写序列号")
pred.init(model_dir={RES文件夹路径}, device=edge.Device.CPU, engine=edge.Engine.PADDLE_FLUID)
pred.infer_image({numpy.ndarray的图片})
pred.close()

    def init(self,
            model_dir,
            device=Device.LOCAL,
            engine=Engine.PADDLE_FLUID,
            config_file='conf.json',
            preprocess_file='preprocess_args.json',
            model_file='model',
            params_file='params',
            graph_file='graph.ncsmodel',
            label_file='label_list.txt',
            device_id=0
            ):
       """
       Args:
           device: Device.CPU
           engine: Engine.PADDLE_FLUID
           model_dir: str
               model dir
           preprocess_file: str
           model_file: str
           params_file: str
           graph_file: str
           label_file: str
           device_id: int

       Raises:
           RuntimeError, IOError
       Returns:
           bool: True if success

       """

pred.init(model_dir=_model_dir, device=edge.Device.CPU, engine=edge.Engine.PADDLE_FLUID, thread_num=1)

    def infer_image(self, img,
                   threshold=0.3,
                   channel_order='HWC',
                   color_format='BGR',
                   data_type='numpy'):
       """

       Args:
           img: np.ndarray or bytes
           threshold: float
               only return result with confidence larger than threshold
           channel_order: string
               channel order HWC or CHW
           color_format: string
               color format order RGB or BGR
           data_type: string
               image data type
           
       Returns:
           list

       """

{
    "index": 736,
    "label": "table",
    "confidence": 0.9
}

{
    "y2": 0.91211,
    "label": "cat",
    "confidence": 1.0,
    "x2": 0.91504,
    "index": 8,
    "y1": 0.12671,
    "x1": 0.21289
}

{
	  "name": "cat",
    "score": 1.0,
    "location": {
    	"left": ..., 
    	"top": ..., 
    	"width": ...,
    	"height": ...,
    },
    "mask": ...
}

{
  {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

    def infer_image(self, sound_binary,
                   threshold=0.3):
       """

       Args:
           sound_binary: sound_binary
           threshold: confidence
           
       Returns:
           list

       """

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]