Pretty cool but seems like there’s a big limitation on this for now “ The pretrained model probably not work for data from another camera sensor. We do not hav

Pretty cool but seems like there’s a big limitation on this for now “ The pretrained model probably not work for data from another camera sensor. We do not have support for other camera data. It also does not work for images after camera ISP, i.e., the JPG or PNG data.” Would be cool to see how they come up with better models that would allow them to overcome the above limitations