02/03/records/001663.json

{
  "sample_id": "001663",
  "status": "annotated",
  "annotation": {
    "sample_id": "001663",
    "task": "target_grounding",
    "stress": {
      "M": [
        "specular_confusion"
      ]
    },
    "instruction": "Locate the green highlighter in the image (the physical object on the table, not any mirror reflection of it). Respond with one axis-aligned bounding box as a single tuple (x_min, y_min, x_max, y_max). Each value must be an integer between 0 and 1000 (permille): x is scaled by image width and y by image height, with 0 at the left/top edge and 1000 at the right/bottom edge. Do not include any text outside the tuple.",
    "rgb": "02/03/media/rgb/001663_rgb.png",
    "gt": {
      "type": "bbox",
      "bbox": {
        "format": "xyxy",
        "space": "permille",
        "box": [
          726,
          409,
          812,
          453
        ]
      }
    }
  }
}