We should have a common interface for basic attributes, bag of words, local feature forests, global features, and even CNN features.
I have been working to create a common class structure for these, in a way where they can be easily merged together to construct larger feature vectors.
The methods of this class will support fit() and transform() so we can work with the sklearn.Pipline structure and explore model parameters easily.
We should have a common interface for basic attributes, bag of words, local feature forests, global features, and even CNN features.
I have been working to create a common class structure for these, in a way where they can be easily merged together to construct larger feature vectors.
The methods of this class will support
fit()andtransform()so we can work with thesklearn.Piplinestructure and explore model parameters easily.