SR-VAE turns widefield fluorescence microscopy images into 2× super-resolved ones. It is a conditional variational autoencoder with a learned, input-conditioned sampler, so the encoder is dropped at inference. It is trained on 20,250 BioSR CCPs pairs and beats bicubic interpolation at all 9 signal levels (+7.3 dB PSNR on 135 test images). The repo
nitin020305/Conditional-Variational-Autoencoder-for-Super-Resolution-of-Fluorescence-Microscopy-: SR-VAE turns widefield fluorescence microscopy images into 2× super-resolved ones. It is a conditional variational autoencoder with a learned, input-conditioned sampler, so the encoder is dropped at inference. It is trained on 20,250 BioSR CCPs pairs and beats bicubic interpolation at all 9 signal levels (+7.3 dB PSNR on 135 test images). The repo
