Implementation of Semantic Pyramid for Image Generation (https://arxiv.org/abs/2003.06221) in PyTorch
Details of model implementation will be different to offical model, as currently official model is not yet released.
> python -m torch.distributed.launch --nproc_per_node=[NUM GPUS] --master_port=[PORT] train.py [Places365 PATH]
First column is source images that vgg features extracted from. Next columns is images generated by conditioned on the specific level of the features. (Conv1 ~ Conv5, FC7, FC8)