This model is trianed by Shenasa.ia cars dataset (which is private) and VCoR (Vehicle Color Recognition) Dataset from kaggle. Images are resized to 256x256 and some classes are omitted due to ...