Detect all interested objects but only showing Bbox for the ones with the full shape not part!

[Auth0rM0rgan]

I have trained my model to detect the different object on my custom dataset, the model detect the object very well and I got a good accuracy, now I want to add some post-processing to not showing bounding box of the objects if there are missing part of it (I want to detect all the possible objects but only showing the ones with the full shape), for example, the model detects correctly the object if, with only 50% of the object in the image or even less with high confidence, I don't want to show the bbox of the object when there is obstacle over the object or when there is only part of the object, Any idea how can I do it?

Thanks a lot!