2

I'm working on object detection using dlib, I was going through python implementation. I tested couple of examples from dlib python examples, specially i worked on train_object_detector.py. this works well. Now I would like to train same data model on CNN based object detector mode But i could not find python implementation for training CNN using python, but there is c++ example. (dnn_mmod_ex.cpp). I think it seems i'm missing something or python implementation is not available? if python implementation is not available, then should i switch to c++ for CNN based object detector training

bommina
  • 307
  • 4
  • 16

1 Answers1

4

Yes, use C++ for CNN training. The dlib DNN tooling is meant to be used from C++ and so uses C++11 features that can't be represented in Python.

Davis King
  • 4,731
  • 1
  • 25
  • 26
  • 1
    Thank you Davis. This was expected and I started working last week c++ cnn mod, But I thought it might be available with dlib 19.7. Still it’s so good, CNN mod is working well. But it took 74 hrs to train examples/faces 4 images in my I7 3.2Ghz with 16Gb RAM, with out GPU. I’m bit worried to train 2000 images on same machine. Can you suggest any GPU or specification to do this kind of training. Thank you – bommina Nov 06 '17 at 13:59
  • 2
    Get a 1080ti. That’s what I’m using and it’s super fast and reasonably priced. – Davis King Nov 06 '17 at 14:29
  • By the way way, I've faced the fact that on 4 GB GRAM face recognition / dlib can proccess 700-800 kB picture files: maximum: larger files couse GPU / out of GRAM eceptions. – Bohdan Feb 01 '22 at 15:37