teammates: Huang Junkai, Huang Yiming, Liu Pak Hin
We leveraged gradient loss and segmentation to improve neural style transfer pipeline.
HRNet.py is imported from another project.
style_transfer.py adds gredient loss.
style_transfer_HRNet.py adds HRNet in front.