A compact model (approx. 1M params.) to hallucinate low resolution (24x24, 16x16 & 8x8) RGB face images with an upscaling factor of 8x.
Use the pre-trained model to hallucinate SR images, or train the model with your dataset from scratch.
aashishrai3799/Super-Resolution-of-Tiny-Face-Images
A compact model to super-resolve low resolution facial images.
MIT