Image Annotation（图像标注(Image Annotation)）

The Image Annotation widget is used to annotate images by drawing rectangles around areas of interest.

The screenshot below shows an example of a configured Image annotation widget in the process of creating an annotation:

Configuration options

Here is a screenshot of the initial state of a newly-added Image Annotation widget alongside its initial configuration panel:

• Input data
• Image source: The image can be displayed from either a media URL or media reference. Currently this widget accepts the following image file types:
  • Media URL: Select a string variable with a valid media URL to render a preview of the media. If referencing a media URL from a dataset, the URL should be in either of the following format: .png, .jpg, .jpeg, .bmp, or .webp,
  • https://{my-foundry-url}/foundry-data-proxy/api/web/dataproxy/datasets/{dataset rid}/transactions/{transaction rid}/{filename}
  • https://{my-foundry-url}/foundry-data-proxy/api/web/dataproxy/datasets/{dataset rid}/views/{branch name}/{filename}.
  • Otherwise, if referencing an external media URL, configure access in your enrollment's Content Security Policy settings.
  • Media reference: Define an object set with a single object and select the media reference typed property to render a preview of the media for that object.
• Current bounding box selection: A string variable that tracks the active bounding box selection coordinate stored as "x1, y1, x2, y2", where (x1, y1) are the pixel coordinates for the upper-left corner of the box, and (x2, y2) are the pixel coordinates for the lower-right corner of the box.
• Active annotation object: The object representing the currently selected annotation.
• Annotations
• Annotation objects: The object set representing the annotations to be superimposed on the image. Currently, up to 1000 annotations per image is supported.
• Bounding box property: The annotation object set's string property that represents the point vector.
• Color property: The annotation object set's string property that represents the color to be drawn as a hex code. This field is optional.
• On annotation create: Enable module builders to configure Workshop events to trigger when an annotation is created by a user.

---

中文翻译

---

图像标注(Image Annotation)

图像标注(Image Annotation) 组件用于通过绘制矩形框来标注图像中的感兴趣区域。

下方截图展示了一个已配置的图像标注(Image Annotation) 组件在创建标注过程中的示例：

配置选项

以下是新添加的图像标注(Image Annotation)组件初始状态及其初始配置面板的截图：

• 输入数据(Input data)
• 图像来源(Image source)： 图像可通过媒体URL或媒体引用(media reference)显示。目前该组件支持以下图像文件类型：
  • 媒体URL(Media URL)： 选择包含有效媒体URL的字符串变量，以渲染媒体预览。如果引用数据集中的媒体URL，该URL应采用以下格式之一：.png、.jpg、.jpeg、.bmp 或 .webp，
  • https://{my-foundry-url}/foundry-data-proxy/api/web/dataproxy/datasets/{dataset rid}/transactions/{transaction rid}/{filename}
  • https://{my-foundry-url}/foundry-data-proxy/api/web/dataproxy/datasets/{dataset rid}/views/{branch name}/{filename}。
  • 否则，如果引用外部媒体URL，请在注册的内容安全策略(Content Security Policy)设置中配置访问权限。
  • 媒体引用(Media reference)： 定义一个包含单个对象的对象集，并选择媒体引用(media reference)类型属性，以渲染该对象的媒体预览。
• 当前边界框选择(Current bounding box selection)： 一个字符串变量，用于跟踪当前选中的边界框坐标，存储格式为"x1, y1, x2, y2"，其中(x1, y1)为框左上角的像素坐标，(x2, y2)为框右下角的像素坐标。
• 当前标注对象(Active annotation object)： 表示当前选中标注的对象。
• 标注(Annotations)
• 标注对象(Annotation objects)： 表示要叠加在图像上的标注的对象集。目前每张图像最多支持1000个标注。
• 边界框属性(Bounding box property)： 标注对象集中的字符串属性，用于表示点向量。
• 颜色属性(Color property)： 标注对象集中的字符串属性，用于表示绘制颜色的十六进制代码。此字段为可选。
• 创建标注时(On annotation create)： 允许模块构建者配置Workshop事件，以便在用户创建标注时触发。