› Robot building › Robot Steps › Cloud AI

Cloud AI

The Cloud AI step provides easy access to AI functionality found in the three cloud service providers: Amazon Web Services, Microsoft Azure, and Google Cloud. The functionality differs slightly between the three, however, you can do the following using each of the providers:

  • Translate text to another language

  • Analyze an image and identify objects in it

The functionality of the step depends on the selected cloud service provider. If the functionality of the cloud service is changed by Amazon, Microsoft, or Google, the Cloud AI step might stop working.

The step depends on API keys to execute the step. These keys must be obtained from the cloud providers and are not part of Kofax RPA.

Analyze images

All three providers can analyze images and return a JSON formatted string with the results.

For supported image formats and requirements, refer to each provider documentation.

Amazon Web Services

Amazon requires two API keys named "Access Key ID" and a corresponding "Secret access key". This API key pair must be granted access to the Amazon Translate service. Refer to Amazon AWS documentation for further details.

Specify a region before working with the Amazon services. The region instructs the service where (geographically) the request should be handled and we recommend to pick a region that is close to where the robot is running. You can find a list of supported regions in the documentation for each service. Amazon may change this list at any time.

The Amazon service has three different image analysis methods:

  • Detect labels: detects objects of various kinds.

  • Detect protective gear: detects protective gear like helmets.

  • Detect faces: detects faces and their features.

Refer to the Amazon Web Services documentation for further details.

Microsoft Azure

To use Azure for image analysis, an endpoint and an API keys are required. Obtain an endpoint and API keys by creating a “Cognitive Services” service with the “Computer Vision” API type.

API keys are different for analyzing images and translating text.

Azure automatically directs requests to the nearest region to the requestor.

The image analysis can be configured to return specific computed details. These can be turned on and off in step options.

Refer to Microsoft Azure documentation for further details.

Google Cloud

To use Google Cloud for image analysis, enable access to "Cloud Vision API" and create an API key. Google automatically routes requests to the data center closest to the requestor.

The image analysis can be configured to return specific computed details. These can be turned on and off in step options.

Refer to Google Cloud documentation for further details.

Analyze text

All three service providers have the option of translating text from one language to another. Azure also provides the option of transliteration of text.

Amazon Web Services

Amazon requires two API keys named "Access Key ID" and a corresponding "Secret access key". This API key pair must be granted access to the Amazon Translate service. Refer to Amazon AWS documentation for further details.

When creating a Cloud AI step, specify a language code for the input.

See the list of supported languages at the following location:

https://docs.aws.amazon.com/translate/latest/dg/what-is.html#what-is-languages

Microsoft Azure

To use Azure for text analysis a location and an API key are required. Obtain a location and an API key by creating a “Cognitive Services” service with the “Translator” API type.

API keys are different for analyzing images and translating text.

The Azure provider can translate and transliterate text. When creating a Cloud AI step, select a language for the input. The supported languages and transliterations are downloaded from Azure at design time.

Refer to Microsoft Azure documentation for further details.

Google Cloud

To use the Google Cloud for text analysis, enable access to "Cloud Translation API" and create an API key.

When creating a Cloud AI step, specify a language code for the input text in the Target Language field.

See the list of currently supported languages at the following location:

https://cloud.google.com/translate/docs/languages

Step properties

The following section provides Cloud AI step properties description.

The number of options and their names depend on the cloud service provider.

Data Type

  • Image: Select this option to process an image.

  • Text: Select this option to analyze a text.

Source

  • Image: Select between Local file, Binary data (variable), RFS, and URL. Once you select the type of the source, specify the path to the file, URL, or select a variable.

  • Text: Type or paste the text to the text field or specify a text variable.

Provider

Select a cloud service provider.

After you select a provider, specify keys and other settings necessary to connect to the service. Refer to each service description above.

Action
Select an action and specify one or more options for the selected action.
Language

  • Image: Select one of the languages supported by the service provider. The output JSON file will contain sections in the selected language.

  • Text: Select a language for the analyzed text. For transliteration, specify the Source Language and Target Script.

Results
Specify a text variable to store the results. For the image, the output text contains JSON markup that you can use to form a JSON document.