VGG16 is pre-trained on ImageNet which is RGB but my current dataset consists of grayscale images. As suggested by some other users here, I tried to input all the grayscale