Some data images do not match the questions and answers in screenqa. Is this reasonable?

[ZhangLin009]

{"image_id": 37374, "image_width": 1080, "image_height": 1920, "question": "What type of Ford car is selected?", "ground_truth": [{"full_answer": "The type of selected Ford car is the C Max 2014.", "ui_elements": [{"text": "C Max", "bounds": [33, 319, 939, 376], "vh_index": 12}, {"text": "2014", "bounds": [33, 483, 939, 540], "vh_index": 15}]}, {"full_answer": "The selected type of "Ford" car is "C Max".", "ui_elements": [{"text": "C Max", "bounds": [33, 319, 939, 376], "vh_index": 12}]}, {"full_answer": "The selected type of Ford car is "C Max".", "ui_elements": [{"text": "C Max", "bounds": [33, 319, 939, 376], "vh_index": 12}]}]}
image_id 37374 in rico is a black and empty image.