AI model in Bitskout is a ready made function to label a content according to defined user parameters. Right now Bitskout supports following content for AI analysis:
Videos (up to 5 min) - not yet supported.
Audio (MP3/WAV) - not yet supported.
Bitskout Image model allows you detect general objects on the image and detect text on the same image.
Generally, you use labels as a way to describe an image without looking at it. For instance, in Facility Management case, the contractor has installed the window, then the labels you will be looking for are "window", "glass", "building" in addition to the location of the image.
It is important to make a clear separation here - default Bitskout AI models are generic models, they cannot detect a window make or assess the quality of window installation or find defects in brick wall. For that you will need to create a Custom Labels model or find a model provider and integrate it to Bitskout.
Creating Image Model Example
Text models are used to analyze text. There are four types of analysis:
Topic Modeling (not yet supported)
You can configure Bitskout Text Model only for one type of analysis at once:
Bitskout used generic NLP models for text and if you need a special model with special terminology (like legal terminology, for instance), then you will need a custom model.