您现在访问的是微软AZURE全球版技术文档网站,若需要访问由世纪互联运营的MICROSOFT AZURE中国区技术文档网站,请访问 https://docs.azure.cn.

计算机视觉中的新增功能What's new in Computer Vision

了解服务中的新增功能。Learn what's new in the service. 这些内容可能包括发布说明、视频、博客文章和其他类型的信息。These items may be release notes, videos, blog posts, and other types of information. 请将本页加入书签,以随时了解该服务的最新信息。Bookmark this page to stay up to date with the service.

2021 年 4 月April 2021

计算机视觉 v3.2 GAComputer Vision v3.2 GA

计算机视觉 API v3.2 现已正式发布,进行了以下更新:The Computer Vision API v3.2 is now generally available with the following updates:

  • 改善了图像标记模型:根据图像中显示的对象、操作和内容分析视觉对象内容并生成相关标记。Improved image tagging model: analyzes visual content and generates relevant tags based on objects, actions and content displayed in the image. 此功能通过标记图像 API 提供。This is available through the Tag Image API. 有关详细信息,请参阅图像分析操作指南概述See the Image Analysis how-to guide and overview to learn more.
  • 更新的内容审查模型:检测是否存在成人内容并提供标志来筛选包含成人、猥亵和血腥视觉内容的图像。Updated content moderation model: detects presence of adult content and provides flags to filter images containing adult, racy and gory visual content. 这可通过分析 API 获得。This is available through the Analyze API. 有关详细信息,请参阅图像分析操作指南概述See the Image Analysis how-to guide and overview to learn more.
  • 用于 73 种语言的 OCR(读取),这些语言包括简体中文和繁体中文、日语、韩语和拉丁语言。OCR (Read) available for 73 languages including Simplified and Traditional Chinese, Japanese, Korean, and Latin languages.
  • OCR (读取) 还可作为本地部署的 Distroless 容器OCR (Read) also available as a Distroless container for on-premise deployment.

2021 年 3 月March 2021

计算机视觉 3.2 公共预览版更新Computer Vision 3.2 Public Preview update

计算机视觉 API 3.2 公共预览版已经更新。The Computer Vision API v3.2 public preview has been updated. 该预览版包含所有计算机视觉功能以及已更新的读取 API 和分析 API。The preview release has all Computer Vision features along with updated Read and Analyze APIs.

2021 年 2 月February 2021

读取 API v3.2 公共预览版(带有对 73 种语言的 OCR 支持)Read API v3.2 Public Preview with OCR support for 73 languages

计算机视觉的读取 API v3.2 公共预览版(可用作云服务和 Docker 容器)包括以下更新:Computer Vision's Read API v3.2 public preview, available as cloud service and Docker container, includes these updates:

  • 用于 73 种语言的 OCR,这些语言包括简体中文和繁体中文、日语、韩语和拉丁语言。OCR for 73 languages including Simplified and Traditional Chinese, Japanese, Korean, and Latin languages.
  • 文本行输出的自然读取顺序(仅限拉丁语言)Natural reading order for the text line output (Latin languages only)
  • 文本行的手写样式分类以及置信度分数(仅限拉丁语言)。Handwriting style classification for text lines along with a confidence score (Latin languages only).
  • 对于多页文档,仅提取所选页面的文本。Extract text only for selected pages for a multi-page document.
  • 可为本地部署用作 Distroless 容器Available as a Distroless container for on-premise deployment.

若要了解详细信息,请参阅读取 API 操作指南See the Read API how-to guide to learn more.

2021 年 1 月January 2021

空间分析容器更新Spatial Analysis container update

已发布提供新功能集的空间分析容器新版本。A new version of the Spatial Analysis container has been released with a new feature set. 借助此 Docker 容器,可分析实时流视频,了解人们与他们在物理环境中的移动之间的空间关系。This Docker container lets you analyze real-time streaming video to understand spatial relationships between people and their movement through physical environments.

  • 现可配置空间分析操作来检测某人是否正戴着口罩等保护性面罩。Spatial Analysis operations can be now configured to detect if a person is wearing a protective face covering such as a mask.
    • 可通过配置 ENABLE_FACE_MASK_CLASSIFIER 参数,为 personcountpersoncrossinglinepersoncrossingpolygon 操作启用口罩分类器。A mask classifier can be enabled for the personcount, personcrossingline and personcrossingpolygon operations by configuring the ENABLE_FACE_MASK_CLASSIFIER parameter.
    • 系统将以元数据的形式返回 face_maskface_noMask 属性,其中有在视频流中检测到的每个人的置信度分数The attributes face_mask and face_noMask will be returned as metadata with confidence score for each person detected in the video stream
  • personcrossingpolygo 操作已得到扩展,可计算一个人在某个区域中的停留时间。The personcrossingpolygon operation has been extended to allow the calculation of the dwell time a person spends in a zone. 可将该操作的区域配置中的 type 参数设置为 zonedwelltime,类型为 personZoneDwellTimeEvent 的新事件将包括 durationMs 字段,该字段填充了该人员在该区域中停留的毫秒数。You can set the type parameter in the Zone configuration for the operation to zonedwelltime and a new event of type personZoneDwellTimeEvent will include the durationMs field populated with the number of milliseconds that the person spent in the zone.
  • 中断性变更:已将 personZoneEvent 事件重命名为 personZoneEnterExitEvent 。Breaking change: The personZoneEvent event has been renamed to personZoneEnterExitEvent. 此事件在某人进入或离开该区域时由 personcrossingpolygon 操作引发,并提供与所穿过区域的编号侧相关的方向信息。This event is raised by the personcrossingpolygon operation when a person enters or exits the zone and provides directional info with the numbered side of the zone that was crossed.
  • 可在所有操作中将视频 URL 作为“专用参数/已模糊处理”提供。Video URL can be provided as "Private Parameter/obfuscated" in all operations. 模糊处理现在是可选操作,仅当 KEYIV 作为环境变量提供时才有效。Obfuscation is optional now and it will only work if KEY and IV are provided as environment variables.
  • 默认情况下,对所有操作启用了校准。Calibration is enabled by default for all operations. 设置 do_calibration: false 可禁用它。Set the do_calibration: false to disable it.
  • 已通过 enable_recalibration 参数增加对自动重新校准的支持(默认禁用),请参阅空间分析操作了解有关详细信息Added support for auto recalibration (by default disabled) via the enable_recalibration parameter, please refer to Spatial Analysis operations for details
  • 照相机校准参数设置为 DETECTOR_NODE_CONFIGCamera calibration parameters to the DETECTOR_NODE_CONFIG. 有关详细信息,请参阅空间分析操作Refer to Spatial Analysis operations for details.

2020 年 10 月October 2020

计算机视觉 API v3.1 GAComputer Vision API v3.1 GA

正式发布的计算机视觉 API 已升级到 v3.1。The Computer Vision API in General Availability has been upgraded to v3.1.

2020 年 9 月September 2020

空间分析容器预览版Spatial Analysis container preview

空间分析容器现提供预览版。The Spatial Analysis container is now in preview. 利用计算机视觉的空间分析功能,你可以分析实时流视频,了解人们与他们在物理环境中的移动之间的空间关系。The Spatial Analysis feature of Computer Vision lets you to analyze real-time streaming video to understand spatial relationships between people and their movement through physical environments. 空间分析是一种可以在本地使用的 Docker 容器。Spatial Analysis is a Docker container you can use on-premises.

读取 API v3.1 公共预览版添加了日语的 OCRRead API v3.1 Public Preview adds OCR for Japanese

计算机视觉的读取 API v3.1 公共预览版添加了以下功能:Computer Vision's Read API v3.1 public preview adds these capabilities:

  • 日语的 OCROCR for Japanese language

  • 对于每个文本行,指示呈现效果是手写体还是打印样式,并随附置信度评分(仅限拉丁语言)。For each text line, indicate whether the appearance is Handwriting or Print style, along with a confidence score (Latin languages only).

  • 对于多页文档,仅提取所选页面或页面范围的文本。For a multi-page document extract text only for selected pages or page range.

  • 此预览版本的读取 API 支持英语、荷兰语、法语、德语、意大利语、日语、葡萄牙语、简体中文和西班牙语。This preview version of the Read API supports English, Dutch, French, German, Italian, Japanese, Portuguese, Simplified Chinese, and Spanish languages.

若要了解详细信息,请参阅读取 API 操作指南See the Read API how-to guide to learn more.

2020 年 7 月July 2020

读取 API v3.1 公共预览版包含简体中文的 OCRRead API v3.1 Public Preview with OCR for Simplified Chinese

计算机视觉的读取 API v3.1 公共预览版添加了对简体中文的支持。Computer Vision's Read API v3.1 public preview adds support for Simplified Chinese.

  • 此预览版本的读取 API 支持英语、荷兰语、法语、德语、意大利语、葡萄牙语、简体中文和西班牙语。This preview version of the Read API supports English, Dutch, French, German, Italian, Portuguese, Simplified Chinese, and Spanish languages.

若要了解详细信息,请参阅读取 API 操作指南See the Read API how-to guide to learn more.

2020 年 5 月May 2020

计算机视觉 API 3.0 版本正式发布,并对读取 API 进行了更新:Computer Vision API v3.0 entered General Availability, with updates to the Read API:

  • 支持英语、荷兰语、法语、德语、意大利语、葡萄牙语和西班牙语Support for English, Dutch, French, German, Italian, Portuguese, and Spanish
  • 准确度改进Improved accuracy
  • 每个已提取单词的置信度分数Confidence score for each extracted word
  • 新输出格式New output format

若要了解详细信息,请参阅 OCR 概述See the OCR overview to learn more.

2020 年 3 月March 2020

2020 年 1 月January 2020

读取 API 3.0 公共预览版Read API 3.0 Public Preview

现在,可以选择使用 Read API 3.0 版从图像中提取印刷体文本或手写文本。You now have the option to use version 3.0 of the Read API to extract printed or handwritten text from images. 与早期版本相比,3.0 版提供了:Compared to earlier versions, 3.0 provides:

  • 准确度改进Improved accuracy
  • 新输出格式New output format
  • 每个已提取单词的置信度分数Confidence score for each extracted word
  • 使用附加的语言参数同时支持西班牙语和英语Support for both Spanish and English languages with the additional language parameter

按照提取文本快速入门,开始使用 3.0 API。Follow an Extract text quickstart to get starting using the 3.0 API.

认知服务更新Cognitive Service updates

认知服务的 Azure 更新公告Azure update announcements for Cognitive Services