Shypd.aiThe largest open index of AI toolsSFTGAN is an image super-resolution tool that recovers realistic texture using deep spatial feature transform. It provides both Torch and PyTorch versions for testing.
Trending
About
SFTGAN is a research-based tool, presented at CVPR 2018, focused on recovering realistic texture in image super-resolution through a deep spatial feature transform. The repository offers simple testing codes in both original Torch and a PyTorch version. For comprehensive training and testing functionalities, users are directed to the BasicSR repository. SFTGAN introduces a Spatial Feature Transform (SFT) layer to efficiently integrate categorical conditions into a CNN network, leveraging semantic segmentation maps as a categorical prior for super-resolution. It also provides access to the OST (Outdoor Scenes) dataset for training and testing, which includes 7 categories of images with rich textures and 300 test images.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
