Image Segmentation Hypothesis

When creating a 2d Segmentation task, you can provide a hypothesis or pre-label attempt.

The format of the hypothesis should match that of the task response. The two required components of the response will be the labelMapping and the annotations.combined.indexedImage parameter.

"hypothesis": {
  "labelMapping": {
    "person": {
      "index": 1
    }
  },
  "annotations": {
    "combined": {
      "indexedImage": "https://example.com/mask.png"
    }
  }
}