Operation op_image_analysis_openai for intelligent image analysis

There is an operation called op_image_analysis_openai that can be launched on a document that has a main binary for intelligent image analysis. The analysis will be stored in the document feature called “openai.image.analysis” and will return whatever you want depending on the context you have specified.

To specify the context, you must go to the configuration of that operation and type in the text box the desired context. For example, imagine that you want to identify the type of defect that a specific appliance has by the attached image and you have a choice field with the following labels and values:

We want the operation to return the value (left column) of the type of damage present and to be stored in the choice field. To do this, the context that would be set could be something like this:

Analyze the possible damage to the device shown in the image (if any) from among the following:

Scratch (internal value “scratch”): marks or scratches on the surface
Dent (internal value “dent”): Surface depression caused by a blow
Crack/Crack (internal value “crack_crack”): crack or opening in which the object is not completely divided
Perforation (internal value “perforation”): Hole produced on the surface, probably by blow or impact against a blunt element.
Rupture (internal value “rupture”): separation of parts of a whole
Deteriorated interior (internal value “interior_deteriorated”): applies to refrigerators or other household appliances with an interior compartment. Refers to any type of deterioration.

Return me as answer only and exclusively the internal value of the type of damage it presents (it can only be one, the clearest one), in case of not presenting it, return “NA”.

Once you have specified the context in which the AI is told to return only the internal value to be taken by the choice, on launching the operation and as specified above, the response will be stored in the document feature “openai.image.analysis”. If you want it to be stored in the choice field, you can create a dynamic expression from the edit of that field to extract the value stored in the “openai.image.analysis” feature such that fil.get_feature_value(“openai.image.analysis”) is extracted in that field:

Once the dynamic expression is added, when the field is extracted it will store what the AI returned with the operation.