Lets say that I have a model that detects Apples, oranges and grapes in an image and I also have another model that detects Jack fruit and Banana in a image. So how do I create a model such that the model can detect Jack fruit, Banana, Apples, oranges and grapes?

My main query is how to combine both the above models?

I searched online and I found out about Ensemble technique, but I think ensemble techniques use the model which predict the same thing.


1 Answer 1


So, an ensemble model is typically describing two or more models looking to predict the same outcomes (classes or values) and then you combine the outputs in some ways.

With a binary classifier, if there were three models you would take the dominant response, best 2 out of three if they did not all agree.

On a continuous value problem you might use the mean value of the three predictions.

In the instance which you are talking about, these are two separate models predicting different things, so you cannot ensemble them because an image passed into each model may return something, an apple and a jack fruit.

This takes you from the realm of simple ML modeling and into the realm of building an artificial intelligence system.

You would either use a pre-made management system, or write software to take incoming images, pass them to each model and capture the output of each model and promote it to whatever that next step is.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.