Hey Vijay,

The model itself can’t magically know what object should be recognized. It’s all down to how you train the model.

In a machine learning example, if you give it enough input examples, it’s going to be better at providing suitable matches for those objects.

So to answer your question – without context (i.e. training) it can’t do it.

How will any model understand objects in drawings if there are no context provided and its very common practice individuals use their own ways of representing them as there are no industry standards. for e.g. you may prefer oval for garden table i may prefer rectangle some one may prefer an circle. With out context how will model classify and help generate anything useful.