Microsoft Computer Vision API入门指南：Cognitive Services实战教程

需积分: 3 34 浏览量更新于2024-07-19 收藏 2.41MB PDF 举报

《微软计算机视觉API精粹：认知服务入门指南》是一本由Alessandro Del Sole编写的书籍，专门介绍Microsoft Computer Vision APIs在Cognitive Services中的应用。该书深入浅出地讲解了如何利用这些高级技术进行图像识别、分析和理解，是开发者理解和实践Microsoft AI解决方案的重要参考。本书涵盖了Microsoft Computer Vision API的核心概念，包括但不限于： 1. **API概述**：介绍了Microsoft Cognitive Services的整体架构，特别是与计算机视觉相关的组件，如API接口、RESTful服务以及如何通过API调用来处理图像数据。 2. **基础功能**：涵盖了基础的图像识别（如物体检测、人脸识别）、图像分类、图像描述生成等内容，让读者快速上手使用API进行基本的图像处理任务。 3. **高级功能**：深入探讨了更复杂的技术，如光学字符识别（OCR）、图像分割、场景理解等，展示了API在实际项目中的强大潜力。 4. **案例研究**：书中提供了丰富的案例分析，展示了如何将Computer Vision APIs应用于电商、社交媒体监控、自动驾驶等领域的具体解决方案。 5. **部署和集成**：讲解了如何将API集成到Web应用程序、移动应用或其他开发平台，以及如何优化性能和管理API使用成本。 6. **安全性与隐私**：强调了在使用计算机视觉API时关于数据安全和隐私保护的最佳实践，确保合规性。 7. **最新版本和更新**：由于版权信息中的ISBN和DOI表明这是一本2018年的出版物，书中可能包含了当时可用的最新API版本及其特性，但可能不包含后来的更新和改进。《微软计算机视觉API精粹》适合希望深入了解并运用这些先进技术的开发人员、数据科学家、机器学习工程师，以及任何对计算机视觉感兴趣的人。它不仅提供理论知识，还为实战操作提供了清晰的指导，是进阶学习和项目开发的实用指南。如果你正在寻找一份全面且易于理解的Microsoft Computer Vision API教程，这本书无疑是你的理想选择。

A. Del Sole, Microsoft Computer Vision APIs Distilled,

https://doi.org/10.1007/978-1-4842-3342-9_2

CHAPTER 2

Getting Started with the

Computer Vision API

Imagine you want to build apps that help people with disabilities to understand what’s

around them and to read papers on their behalf.

Now imagine you want to build apps that help kids learn about the world from

pictures, making sure that adult content is excluded. Then, imagine you want to build

apps that help people learn a foreign language by providing natural language sentences

that describe an image. Finally, imagine you work on a police force and want to build a

custom solution that helps your department identify criminals based on images.

These are only a few examples of how artificial intelligence could help solve a

number of problems, but they are enough to make you understand the purpose of the

Computer Vision API. Combined with other Cognitive Services APIs, you have infinite

opportunities. This chapter first describes how the Computer Vision API works, and then

it describes how to configure your Azure subscription and expose a Computer Vision API

endpoint that can be consumed by any application.

Understanding the Computer Vision API

As well as with every other cognitive service, the Computer Vision API has its own

portal; you can reach it at http://azure.microsoft.com/en-us/services/cognitive-

services/computer-vision/. Here you will find shortcuts to the documentation and

examples, but I will want focus on the API reference available at http://bit.ly/2sBtryy.

When you open this page, you will see a list of available operations that you can execute

against images.

Technically speaking, with the Computer Vision API, you invoke a RESTful service

by uploading an image or pointing to an existing image URL and sending GET and POST

HTTP requests depending on the type of analysis you want to execute against the image.

The Computer Vision service will return a JSON response that contains the analysis

results. The service URL varies depending on your closest region, and an active Azure

subscription is required to activate the keys you’ll use in the HTTP requests. You will

configure your Azure subscription in the next section; for now focus on the available

types of analysis you can perform on images and their corresponding HTTP requests,

described in Table2-1.

Chapter 2 ■ GettinG Started with the Computer ViSion api

In this particular case, you will get a JSON response that contains an array of tags,

each with a name and confidence percentage. You also get additional information, such

as the request’s unique identifier and the image size. Other operations provided by

the Computer Vision API can be invoked in a similar way, changing the endpoint and

providing the required parameters.

Whether you work with a debugging tool like Postman or with specialized classes in

your favorite programming language, you will need to supply the information described

earlier. In the next chapter, you will see how to perform all the available operations

from C# with the HttpClient class and how to parse the JSON response in code. In the

meantime, you can try the API testing console included in the Computer Vision API

portal (available at http://bit.ly/2sBtryy). To accomplish this, select an operation

on the left and then click the button that represents your Azure region near the “Open

API testing console” box. At this point, a series of text boxes will appear, and you will find

guidance of how to fill them out and create a POST request to the service. You will then be

able to see the JSON response if the operation succeeds, or you will see an error message

if it fails.

Handling the HTTP Response

Like any other RESTful service, Computer Vision returns an HTTP code and a description

that allows you to understand whether an operation succeeded or not, and the reason.

Table2-2 summarizes the most common HTTP status codes that you might get back

when working with Computer Vision.

Table 2-2. Computer Vision Status Codes

HTTP Status Code Description

200 (Success) The requested operation completed successfully,

and an analysis result was returned as JSON.

400 (Bad Request) The requested operation failed with one of the

following self-explanatory errors: InvalidImageUrl,

InvalidImageFormat, InvalidImageSize,

NotSupportedImage.

415 (Unsupported Media Type) The requested operation failed because the supplied

Content-Type header does not match the image

content.

500 (Internal Server Error) The requested operation failed with one of the

following self-explanatory errors: FailedToProcess,

Timeout, InternalServerError.

401 (Unauthorized) The requested operation could not be executed

because an invalid subscription key was supplied.

剩余97页未读，继续阅读

yinkaisheng-nj

粉丝: 762
资源: 6231

Microsoft Computer Vision API入门指南：Cognitive Services实战教程

Microsoft Computer Vision APIs Distilled Getting Started with Cognitive epub

consider using a system image with google apis to enable testing with google play services

使用android studio 运行apk提示consider using a system image with google apis to enable testing with google play services

aspose-pdf 实现 PDF 转 word

使用gateway报Error creating bean with name 'apis' defined in class path resource

mmdeploy - error - `mmdeploy.apis.pytorch2onnx.torch2onnx` with call id: 0 f

implement the shipping REST APIs with spring boot + sping mvc + jpa/batis + mysql + jwt

from ..mmpose.apis.inference import inference_topdown ImportError: attempted relative import with no known parent package

Eolink Apikit

proxy: { "/apis": { target: "http://localhost:3000/apis",//接口服务器地址 changeOrigin: true,//是否跨域 pathRewrite: { "^/apis": "" } } }这段代码什么意思

最新资源