ThingWorx Flow > Connectors Overview > System Connectors > Azure > Computer Vision
Computer Vision
Azure's Computer Vision service provides developers with access to advanced algorithms that process images and return information. To analyze an image, you can either upload an image or specify an image URL. The processing algorithms can analyze the content of images in different ways, depending on various visual features. For example, Computer Vision can find all the human faces in an image.
The Computer Vision action supports the images that meet the following requirements:
Format is JPEG, PNG, GIF, or BMP.
File size is less than 4MB.
Dimensions are between 50 x 50 pixels and 4200 x 4200 pixels.
Size is no larger than 10 megapixels.
Use the Computer Vision action to analyze an image, analyze an image by domain, describe an image, and detect and provide insights about the visual features and characteristics of an image.
Using the Computer Vision Action
To use this action in your workflow, connect it to a ThingWorx Flow. To connect to the flow, do the following:
1. Drag the Computer Vision action under the Azure connector to the canvas, place the pointer on the action, and then click or double-click the action. The Computer Vision action window opens.
2. Edit the label name, if needed. By default, the label name is same as the action name.
3. To add a new authorization, refer to the section Authorize Azure in the topic Azure connector.
If you previously added an authorization for Azure, select an authorization from the list.
4. Select the resource group defined in the subscription.
5. In the Computer Vision Account field, select the cognitive service account.
6. In the Provide Image by list, select an option:
Upload File—Specify the path of the file stored in the local storage of the flow engine that you want to upload.
URL—Specify the publicly accessible Image URL.
7. Select a computer vision operation from the Select Specific Service list. See a description of each operation in the sections that follow.
Analyze Image
Analyze Image extracts a rich set of visual features based on the image content. Enter the details that follow:
Visual Features—Click Add, and then select what type of visual features to include:
Categories—Categorizes image content according to a taxonomy defined in documentation.
Description—Describes the image content with a complete English sentence.
Color—Determines the accent color, dominant color, and whether an image is black and white.
Tags—Tags the image with a detailed list of words related to the image content.
Faces—Detects if faces are present in the image.
Image Type—Detects if an image is a clip art or a line drawing.
Adult—Detects if the image is pornographic in nature or if the image depicts sexually suggestive content.
Objects—Detects various objects within an image, including the approximate location.
Details—Click Add and then select which domain-specific details to include:
Celebrities—Identifies celebrities if detected in the image
Landmarks—Identifies notable landmarks in the image.
Language—Select the language in which to generate an output, and then click Done. The default is English.
Analyze Image by Domain
Analyze Image by Domain recognizes content within an image by applying a domain-specific model.
Model—Domain-specific content to recognize.
Language—Select the language in which to generate an output, and then click Done. The default is English.
Describe Image
Describe Image generates a description of an image in human readable language with complete sentences. Enter the details that follow:
Max Candidates—Enter the maximum number of candidate descriptions to return.
Language—Select the language in which to generate an output, and then click Done. The default is English.
Detect Objects
Detect Objects performs object detection on the specified image.
Generate Thumbnail
Generate Thumbnail generates a thumbnail image with the user-specified width and height. Enter the details that follow:
Width—Width of the thumbnail, in pixels between 1 and 1024. Recommended minimum of 50.
Height—Height of the thumbnail, in pixels between 1 and 1024. Recommended minimum of 50.
Smart Cropping—Select true for smart cropping or select false, and then click Done.
Extract printed text (OCR)
Optical Character Recognition (OCR) detects text in an image and extracts the recognized characters into a machine-usable character stream. Enter the details that follow:
Language—The BCP-47 language code of the text to be detected in the image. The default is Unknown.
Detect Orientation—Detects the text orientation in the image. Select true to detect the image orientation and to correct it before further processing or select false, and then click Done. For example, if the image is upside-down, the image is corrected before further processing.
Extract Handwritten text
Extract Handwritten text extracts handwritten text from an image.
Tag an Image
Tag an Image generates a list of words, or tags, that are relevant to the content of the supplied image. For Language, select the language in which to generate output, and then click Done. The default is English.
Get Area of Interest
Get Area of Interest returns a bounding box around the most important area of the image.
Output Schema
The Computer Vision action displays an output schema as shown in the figure that follows if you select the Extract Handwritten text service.