Model suggestions for pixelated enhancement of QR codes

I’ve been trying to run models for deblurring and enhancing blurry and distorted qr code images via Keras. However, none of them have been able to return the desired output despite being trained on customized QR code datasets (10,000 images) with variations to the models parameters.

Are there any models which can provide an accurate result with respect to image enhancement of grayscaled images in general that could help with this task?