Azure Computer Vision

For the most up-to-date documentation, please visit docs.brix.ch

The Azure Computer Vision is part of the Microsoft Azure Cognitive Services and provides the following functionalities: Automated tagging, captions, background/foreground/accent color and object recognition. The service can be triggered via context-menu or automatically. Currently only tagging in English is fully supported (only some of the content can be received in another language), but with the Auto Translator plugin, tags can be translated automatically. If English is available, then tags are added in English and the default system language (if it is not English) because the default system language is required, if English isn't available then only the default system language is set (like this tags cannot be translated with the Auto Translator).

Image requirements (therefore large preview (default) is recommended):

  • max 4MB
  • at least 50 x 50 pixels
  • png, jpg, gif, bmp

Properties

To be configured in {home}/appserver/conf/custom.properties

azureComputerVision.license

type: string, required: yes, default: -

The license key for the plugin (product: azureComputerVision), provided by brix.

azureComputerVision.allowedUserGroupIds

type: comma-separated list of user group ids, required: no, default: -

The user group ids which are allowed to use this extension (super admins are always entitled).

azureComputerVision.threads

type: integer, required: no, default: 10

The number of threads in the thread pool, that are processing the asset changes and initiating the automated tagging, has to be >= 2.

azureComputerVision.debounceTimeInSeconds

type: integer, required: no, default: 10

The debounce time in seconds. E.g. 10 means that for 5s all changes that happen on an asset are collected and then processed at once. This has the advantage that if 10 relevant changes happen one after another within an interval of 5s, cognitive services is still only asked once and not 10 times.

azureComputerVision.downloadFormatId

type: integer, required: no, default: large preview (recommended)

Specify a download format id here if not the large preview should be used for the service.

azureComputerVision.url

type: url, required: yes, default: https://westeurope.api.cognitive.microsoft.com/

API-URL for the Computer Vision service.

azureComputerVision.key

type: string, required: yes, default: -

The Computer Vision key.

azureComputerVision.tagsInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the tags. Required to be able to use this feature.

azureComputerVision.tagsThreshold

type: double, required: no, default: 0

Only tags with a certainty (a number between 0 and 1) >= the threshold will be added (this only works for the normal tags and not for those tags sent with the image description).

azureComputerVision.tagsFromDescriptionInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the tags from the description. Required to be able to use this feature.

azureComputerVision.captionInfoFieldIds

type: list of integer, required: no, default: -

The ids of the (localized) text (area) information field for the caption. All types of text information fields are supported. Required to be able to use this feature.

azureComputerVision.backgroundColorInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the dominant background color. Required to be able to use this feature.

azureComputerVision.foregroundColorInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the dominant foreground color. Required to be able to use this feature.

azureComputerVision.dominantColorsInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the dominant colors. Required to be able to use this feature.

azureComputerVision.accentColorInfoFieldId

type: integer, required: no, default: -

The id of the text (area) information field for the accent color (web-format). All types of text information fields are supported. Required to be able to use this feature.

azureComputerVision.blackAndWhiteInfoFieldId

type: integer, required: no, default: -

The id of the checkbox information field for the black and white property. Required to be able to use this feature.

Since v1.0.1 all the information from checkboxes can be collected in a node referencing field. The syntax is <information field id>:<node id>. So instead of setting the checkbox the specified node is added to or removed from the node referencing field.

azureComputerVision.adultContentInfoFieldId

type: integer, required: no, default: -

The id of the checkbox information field for the adult content property. Required to be able to use this feature.

Since v1.0.1 all the information from checkboxes can be collected in a node referencing field. The syntax is <information field id>:<node id>. So instead of setting the checkbox the specified node is added to or removed from the node referencing field.

azureComputerVision.adultScoreInfoFieldId

type: integer, required: no, default: -

The id of the number (score shown as percentage 0 to 100) or double (actual score between 0 and 1) information field for the adult score. Required to be able to use this feature.

azureComputerVision.goryInfoFieldId

type: integer, required: no, default: -

The id of the checkbox information field for the gory property. Required to be able to use this feature.

Since v1.0.1 all the information from checkboxes can be collected in a node referencing field. The syntax is <information field id>:<node id>. So instead of setting the checkbox the specified node is added to or removed from the node referencing field.

azureComputerVision.goryScoreInfoFieldId

type: integer, required: no, default: -

The id of the number (score shown as percentage 0 to 100) or double (actual score between 0 and 1) information field for the gory score. Required to be able to use this feature.

azureComputerVision.racyInfoFieldId

type: integer, required: no, default: -

The id of the checkbox information field for the racy property. Required to be able to use this feature.

Since v1.0.1 all the information from checkboxes can be collected in a node referencing field. The syntax is <information field id>:<node id>. So instead of setting the checkbox the specified node is added to or removed from the node referencing field.

azureComputerVision.racyScoreInfoFieldId

type: integer, required: no, default: -

The id of the number (score shown as percentage 0 to 100) or double (actual score between 0 and 1) information field for the racy score. Required to be able to use this feature.

azureComputerVision.categoriesInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the categories. Required to be able to use this feature.

azureComputerVision.categoriesThreshold

type: double, required: no, default: 0

A value between 0 and 1. Only categories above this threshold are accepted.

azureComputerVision.brandsInfoFieldId

type: integer, required: no, default: -

The id of the node reference information field for the brands. Required to be able to use this feature.

azureComputerVision.objectsInfoFieldId

type: integer, required: no, default: -

The id of the text area information field for the objects (JSON). Required to be able to use this feature.

azureComputerVision.facesInfoFieldId

type: integer, required: no, default: -

The id of the text area information field for the faces (JSON). Required to be able to use this feature.

azureComputerVision.contextMenu

type: boolean, required: no, default: true

Whether the context menu should be available (for the allowed user groups) or not. Restart required.

azureComputerVision.automate

type: boolean, required: no, default: false

Whether images should be tagged automatically or not. Restart required.

type: string, required: no, default: documents=false&videos=false&audios=false&others=false (all images)

Search expression, only tag images within the scope (for automation and initialize task).

azureComputerVision.search.userId

type: integer, required: no, default: api-user

User id of the user to perform the search with (only assets visible to this user will be found).

azureComputerVision.tags.partition.threshold

type: integer, required: no, default: 100

Nodes are partitioned if the number of children in a node (which are not intermediate nodes) becomes greater than the threshold. Like this the GUI doesn't freeze if you try to open the tag tree. A number <= 0 prevents the partitioning.

Installation

  1. Get a license from brix
  2. Create a subscription for Microsoft Azure Cognitive Services > Computer Vision
  3. Select Computer Vision and go to Keys and Endpoints to get the API key
  4. Create the information fields below
  5. Add the information and configuration to the custom.properties file, put the jar file into the lib folder and restart the CELUM app server

The information fields below have to be added to all the asset types which should be available for the Azure Computer Vision extension. Each information field corresponds to one of the functionalities, it is possible to add only some of them and ignore the ones which are of no importance. The fields can be added to an existing or a new fieldset. Tags i.e. nodes will be created inside the root node for the corresponding node referencing information field.

    <nodereference id="?" name="azure_computer_vision_tags">
        <labels>
            <label lang="en">Tags</label>
            <label lang="de">Tags</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="-1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <nodereference id="?" name="azure_computer_vision_tags_from_description">
        <labels>
            <label lang="en">Tags from description</label>
            <label lang="de">Tags aus Beschreibung</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="-1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <!-- it is possible to have more than one caption field, so that if there is more than
    proposed caption, the 2nd best is saved in the 2nd field and so on -->
    <text id="?" name="azure_computer_vision_caption_1">
        <labels>
            <label lang="en">Caption 1</label>
            <label lang="de">Beschriftung 1</label>
        </labels>
    </text>

    <!-- it is recommended to use the same root node as for foreground color and dominant colors -->
    <nodereference id="?" name="azure_computer_vision_background_color">
        <labels>
            <label lang="en">Dominant background color</label>
            <label lang="de">Vorherrschende Hintergrundfarbe</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <!-- it is recommended to use the same root node as for background color and dominant colors -->
    <nodereference id="?" name="azure_computer_vision_foreground_color">
        <labels>
            <label lang="en">Dominant foreground color</label>
            <label lang="de">Vorherrschende Vordergrundfarbe</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <!-- it is recommended to use the same root node as for background color and foreground color -->
    <nodereference id="?" name="azure_computer_vision_dominant_colors">
        <labels>
            <label lang="en">Dominant colors</label>
            <label lang="de">Vorherrschende Farben</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="-1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <text id="?" name="azure_computer_vision_accent_color">
        <labels>
            <label lang="en">Accent color</label>
            <label lang="de">Akzentfarbe</label>
        </labels>
    </text>      

    <checkbox id="?" name="azure_computer_vision_black_and_white">
        <labels>
            <label lang="en">Black and white</label>
            <label lang="de">Schwarzweiss</label>
        </labels>
    </checkbox>

    <checkbox id="?" name="azure_computer_vision_adult_content">
        <labels>
            <label lang="en">Adult content</label>
            <label lang="de">Inhalt für Erwachsene</label>
        </labels>
    </checkbox>

    <double id="?" name="azure_computer_vision_adult_score">
        <labels>
            <label lang="en">Adult score</label>
            <label lang="de">Erwachsenen-Wertung</label>
        </labels>
    </double>

    <checkbox id="?" name="azure_computer_vision_gory">
        <labels>
            <label lang="en">Gory</label>
            <label lang="de">Blutig</label>
        </labels>
    </checkbox>

    <double id="?" name="azure_computer_vision_gory_score">
        <labels>
            <label lang="en">Gory score</label>
            <label lang="de">Blutig-Wertung</label>
        </labels>
    </double>

    <checkbox id="?" name="azure_computer_vision_racy">
        <labels>
            <label lang="en">Racy</label>
            <label lang="de">Gewagt</label>
        </labels>
    </checkbox>

    <double id="?" name="azure_computer_vision_racy_score">
        <labels>
            <label lang="en">Racy score</label>
            <label lang="de">Gewagt-Wertung</label>
        </labels>
    </double>

    <nodereference id="?" name="azure_computer_vision_categories">
        <labels>
            <label lang="en">Categories</label>
            <label lang="de">Karegorien</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="-1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <nodereference id="?" name="azure_computer_vision_brands">
        <labels>
            <label lang="en">Brands</label>
            <label lang="de">Marken</label>
        </labels>
        <!-- create a root node for the keywords and enter its id below -->
        <rootnode maxSelections="-1" showRoot="false" showNodeTree="false">?</rootnode>
    </nodereference>

    <!-- not recommended, unless analyzed/processed by machines -->
    <textarea id="?" name="azure_computer_vision_objects">
        <labels>
            <label lang="en">Objects</label>
            <label lang="de">Objekte</label>
        </labels>
    </textarea>

    <!-- not recommended, unless analyzed/processed by machines -->
    <textarea id="?" name="azure_computer_vision_faces">
        <labels>
            <label lang="en">Faces</label>
            <label lang="de">Gesichter</label>
        </labels>
    </textarea> 

Compatibility Matrix

Azure Computer Vision CELUM
1.0.0 5.13.4 (tested with 6.4)
1.0.1 5.13.4 (tested with 6.4)

Release Notes

1.0.0

Released 2020-10-23

Initial version

1.0.1

Released 2020-10-27

Added possibility to collect boolean fields (checkboxes) in a single node referencing field