search

stac-extensions / ml-model

An Item and Collection extension to describe machine learning (ML) models that operate on Earth observation data.
Apache License 2.0
37 stars 0 forks source link

"Flatten" Training Environment Object #6

Closed m-mohr closed 2 years ago

m-mohr commented 2 years ago

A best practice in STAC has been to put scalar values into properties as much as possible. This simplifies viewing and searching implementation. The Training Environment Object just consists of two string fields so it seems that they could easily be flattened into two top-level properties.

That means the example could be for example:

  "properties": {
    "datetime": "2020-12-11T22:38:32Z",
    "ml-model:learning_approach": "supervised",
    "ml-model:prediction_type": "object-detection",
    "ml-model:architecture": "RCNN",
    "ml-model:training-processor-type": "gpu",
    "ml-model:training-os": "linux"
  },

instead of 

  "properties": {
    "datetime": "2020-12-11T22:38:32Z",
    "ml-model:learning_approach": "supervised",
    "ml-model:prediction_type": "object-detection",
    "ml-model:architecture": "RCNN",
    "ml-model:training-environment": {
      "processor-type": "gpu",
      "operating-system": "linux"
    }
  },