Open adamjstewart opened 1 year ago
Thanks! I think a lot of the goals are similar, although I'm not familiar enough with TorchGeo to know its long term goals and vision. I don't imagine Moonshine having nearly the same level of capability as TorchGeo, and indeed we may use that library for some of the functions.
Currently I think Moonshine's focus will be growing its library of pretrained models and also just trying to understand user issues/use cases. I'm just one dude right now working on this part time, so I doubt Moonshine will be match TorchGeo's breadth or capability. Long term I'd be aiming to work towards a few powerful models instead of a broad array of options, but maybe that's not actually what people want.
If you'd be interested in chatting at all I'm happy to!
Long term goals and vision of TorchGeo is to do all of the things that torchvision does but for satellite imagery and other geospatial data. This includes pre-trained models for as many satellite imagery providers, SSL techniques, and architectures as possible.
We would love for the folks at Moonshine to use TorchGeo, just let us know if there's anything missing. Of course, you can still write your own library, or build it on top of TorchGeo. Just want to make sure y'all are aware of TorchGeo before reinventing the wheel 😄
Sounds awesome! I'd love to chat and get a sense of where I might be able to contribute or collaborate -- I sent you a message on LinkedIn with my personal email.
Just wanted to say thanks for releasing these pre-trained models! There are very few libraries out there that offer models pre-trained on remote sensing imagery, so there's definitely a big need for this.
I was wondering how the goals of your project compare to TorchGeo? We already have ~10 pre-trained models and are planning to expand that to ~100 in the coming year. If you're interested in contributing, we would love to add your pre-trained models as well.