Below are acceptance criteria for #2 the new version of the ML Model Extension. After all these items are done I think #2 is ready to merge and we can publish a new version 2 release to https://github.com/stac-extensions/ml-model/issues . I'm working on most of these, let me know if you think other acceptance criteria should be included cc @fmigneault
[x] Incorporate feedback from @fmariv and others at @earthpulse
I had a meeting with @fmariv, they're keen to review the extension, provide feedback and see if we can align it more with general ML use cases. They're working on supporting the STAC extension ecosystem for ML as a part of the EOTDL initiative
this will involve segregating flat fields from objects that are not meant to be searched on and are instead used for inference model loading, model input and output processing, and documenting in detail the accelerator and runtime details needed to run the model
[x] #4 by creating output objects for common detection tasks. At a minimum, start with single label classification, semantic segmentation, object detection. This will help resolve common ambiguities when interpreting model outputs, like the bbox coordinate ordering for object detection. EDIT this can be left for after the v2 release
[ ] resolve #3, deciding how to refer to model extension metadata. within it's own STAC item/collection json or are these fields composed with common metadata in STAC json representing a spatiotemporal asset?
:rocket: Feature Request
originally posted on https://github.com/crim-ca/dlm-extension/issues/7
Below are acceptance criteria for #2 the new version of the ML Model Extension. After all these items are done I think #2 is ready to merge and we can publish a new version 2 release to https://github.com/stac-extensions/ml-model/issues . I'm working on most of these, let me know if you think other acceptance criteria should be included cc @fmigneault
#4 by creating output objects for common detection tasks. At a minimum, start with single label classification, semantic segmentation, object detection. This will help resolve common ambiguities when interpreting model outputs, like the bbox coordinate ordering for object detection.EDIT this can be left for after the v2 releaseend_datetime: nullnot supported by STAC Core spec (https://github.com/radiantearth/stac-spec/issues/1268)