基于MindSpore Serving部署推理服务

最新推荐文章于 2022-11-24 00:26:29 发布

VIP文章 BaldheadedM

最新推荐文章于 2022-11-24 00:26:29 发布

阅读量291

点赞数 1

本文链接：https://blog.csdn.net/BaldheadedM/article/details/126297086

版权

今天要分享的是服务化部署框架（MindSpore Serving）

具体要实现的就是一个可以在线识图的页面。

1.MindSpore Serving安装

MindSpore Serving目前只能通过指定whl包安装

pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/1.7.0/Serving/x86_64/mindspore_serving-1.7.0-cp39-cp39-linux_x86_64.whl

2.MindSpore Serving使用

快速入门的例子有点不够形象化

于是去github的代码仓里面看看样例都有哪些

serving: A lightweight and high-performance service module that helps MindSpore developers efficiently deploy online inference services in the production environment. - Gitee.com

需要没有GPU，就拿resnet50来跑跑看，cifa10的数据集也不大，cpu下训练也不费事。

如下是相关模型的训练代码：

from mindspore.train import Model
from mindvision.dataset import Cifar10
from mindvision.engine.callback import ValAccMonitor
from mindvision.classification.models.classifiers import BaseClassifier
from mindvision.classification.models.head import DenseHead
from mindvision.classification.models.neck import GlobalAvgPooling
from mindvision.classification.utils.model_urls import model_urls
from mindvision.utils.load_pretrained_model import LoadPretrainedModel
from typing import Type, Union, List, Optional
from mindvision.classification.models.blocks import ConvNormActivation
from mindspore import nn


class ResidualBlockBase(nn.Cell):
    expansion: int = 1  # 最后一个卷积核数量与第一个卷积核数量相等

    def __init__(self, in_channel: int, out_channel: int,
                 stride: int = 1, norm: Optional[nn.Cell] = None,
                 down_sample: Optional[nn.Cell] = None) -> None:
        super(ResidualBlockBase, self).__init__()
        if not norm:
            norm = nn.BatchNorm2d

        self.conv1 = ConvNormActivation(in_channel, out_channel,
                                        kernel_size=3, stride=stride, norm=norm)
        self.conv2 = ConvNormActivation(out_channel, out_channel,
                                        kernel_size=3, norm=norm, activation=None)
        self.relu = nn.ReLU()
        self.down_sample = down_sample

    def construct(self, x):
        """ResidualBlockBase construct."""
        identity = x  # shortcuts分支

        out = self.conv1(x)  # 主分支第一层：3*3卷积层
        out = self.conv2(out)  # 主分支第二层：3*3卷积层

        if self.down_sample:
            identity = self.down_sample(x)
        out += identity  # 输出为主分支与shortcuts之和
        out = self.relu(out)

        return out
class ResidualBlock(nn.Cell):
    expansion = 4  # 最后一个卷积核的数量是第一个卷积核数量的4倍

    def __init__(self, in_channel: int, out_