- modifications to requirements

cam output will be saved in folders

-made default 0 in configs

-fixed issue in test.py

added downscaling cnn option

-fixed issue in training multilabel

-added draw activations

-added best threshold finding

-removed unnecessary code in generator

-added effieicient net import in loading

- added handling for single label classification

-added tensorflow gpu

-added efficient net support

-added generators to make training faster

-handle for extra spaces in csv image names