Created by: lidanqing-intel

In mkldnn_post_training_strategy.py, MKLDNNPostTrainingQuantStrategy only support image classifications. object detection models data reader should be added later.

[removed TODO] slim/quantization/mkldnn_post_training_strategy.py:86:
# TODO (Intel) Remove limits that MKLDNNPostTrainingQuantStrategy only for image classifications’ input issue