Here I have shown how you can ensemble multiple TensorFlow models together
I am ensembling 2 models together by utilizing a third simple model. Data source for given matrix is flower data set in keggle here is the link for the data set https://www.kaggle.com/moltean/fruits
However I have not taken entire dataset as I was doing it on collab with limited resources.
Results are quite interesting as we are getting better accuracy both on training and testing dataset with cross validation.
I have used a simple model in front of two large Transfer Learning models and results are quite promising, However it is quite small improvement in class wise accuracies but better than any single model working alone.
I do not recommend it, if you are going for edge deployment but if for batch inference, go ahead.
Let me know your experience after using these models here [email protected]