Feature Scaling for SVM

asked 2018-07-29 16:52:13 -0500


I am using opencv for classification problem. It's audio classification.

I need to use OpenCV->ML->SVM for it in c++. The reason to use opencv is to avoid compiling/installing another library on hardware which has limited space left for new library to be added.

One of the important requirement for svm is to do feature scaling as mentioned in libsvm guide. Sklearn comes with bunch of scalers, that can be trained using training data and can be stored as pickle object to use again to scale data to predict with trained model while prediction.

My questions are:

  1. What can I use for feature scaling (something similar to stadardscaler or minamaxscaler in sklearn) in opencv c++?
  2. How can I use same scaler while prediction (e.g. is there anyway to store a scaler object same as trained model)?

Thanks, Navdeep

edit retag flag offensive close merge delete