Hi All,
Is there any way to suplement a custom object detection model with rules to assist with predictions i.e. some way to tell the model that the object will always be red and a rectangle for example, I guess I’m talking about supplementing machine vision with LLM.
Does anything like that exist yet?
Thanks
Ben
Hi @Ben_Smith,
I’m not sure about supplementing model with such rules (although you could try to ensure your dataset only contains objects that meet your characteristics, i.e. only red rectangles, and maybe consider adding other objects like green rectangles and labelling them accordingly)
However, you can definitely use workflows to filter out detections based on your characteristics. So instead of training model to only detect red rectangles, you could train the model to detect all types of rectangles and then build a workflow with blocks that filter out those rectangles you are not interested in (i.e. using dominant color block).
Hope this helps,
Grzegorz
This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.