0

I was wondering in what cases using C++ should i use OpenCV's ML libraries CvSVM or CvANN_MLP instead of building them with tensorflow ? which one is better and in what cases ?

JavaBeg2016
  • 93
  • 2
  • 7

1 Answers1

2

If you specifically want to target 'handwritten recognition,' or similar pattern recognition outside of image analysis, neural networks using Tensor Flow is the way to go. OpenCV, on the other hand, is designed specifically for computer vision (essentially high-level image analysis to interpret certain characteristics of digital images).

If you are a beginner in ML, I would highly recommend starting out with TensorFlow as a more broad introduction to ML and specifically to neural networks. There are many great and easy-to-follow resources to get started.

Jonathan H.
  • 939
  • 1
  • 7
  • 21
  • thanks! but this resources isn't written in python ? isn't there any for c++ ? – JavaBeg2016 Jan 03 '17 at 02:43
  • @JavaBeg2016 unfortunately, due to C++ being a less common 'beginner' language, it is more difficult to find basic, descriptive tutorials. You will probably find the Python tutorials helpful regardless, as it will be very similar to the C++ API, and the theory behind both language implementations of TensorFlow will be identical. I recommend checking out the following SO post http://stackoverflow.com/questions/33620794/how-to-build-and-use-google-tensorflow-c-api – Jonathan H. Jan 03 '17 at 02:49
  • thanks! i'll check it. – JavaBeg2016 Jan 03 '17 at 02:52