Adapt model to load 512x512 images from s3 bucket

Clay-foundation / model

The Clay Foundation Model (in development)

https://clay-foundation.github.io/model/

Apache License 2.0

299 stars 38 forks source link

Adapt model to load 512x512 images from s3 bucket #85

Closed weiji14 closed 9 months ago

weiji14 commented 9 months ago

Modify the ViT MAE model to accept input images of size 512x512 pixel after #78. Also making a few small enhancements to the datapipe.

TODO:

[x] Change hardcoded image_size from 256 to 512, patch_size from 32 to 64.
[x] Allow loading of GeoTIFF files directly from s3 bucket instead of local drive
[x] Optimize data loading using sharding filter

weiji14 commented 9 months ago

Gonna merge this PR directly, as it consists of mostly minor tweaks (which I've accumulated over the past few weeks). Some of the options (e.g. patch_size) can be changed later, but thought it'd be good to have a datapipe/model that works with the 512x512 images soon-ish.