Jacob Prince-Bieker
Jacob Prince-Bieker
One possible metric is the Structural Similarity Metric (SSIM), there is a PyTorch differentiable version here that I'm going to try out https://github.com/VainF/pytorch-msssim
Could also try using FID, which is correlated to human perception of visual quality, since we want these generated satellite images to look "natural" https://github.com/mseitzer/pytorch-fid
Working on adding some things from this review paper: https://arxiv.org/pdf/2004.05214.pdf one idea that seems promising for creating a model that can be used as an ensemble, is for the generation...
LPIPS is another one too https://github.com/S-aiueo32/lpips-pytorch
Try it with only calculating the loss on parts of the image stack where things have moved over a certain amount https://github.com/openclimatefix/satflow/issues/88#issuecomment-910233923
Newer loss functions are being put in https://github.com/openclimatefix/nowcasting_utils to share between predict_pv_yield and SatFlow
Hey, sorry for the delay, I just missed this issue. I wouldn't load it all into ram at once. For training, we lazily load the data we need from either...
Hi, I am not sure, I haven't tried the SwinTransformer for these, and we haven't reimplemented the WeatherMesh model yet either. Good to know about its potential downsides though!
Yep! That would be great
> Hey @jacobbieker, > I've added code to fetch hourly data from different NWPs via the OpenMeteo API. Any quick thoughts on how we can compare these NWPs effectively? Also,...