AI-102 Question 118
Select 3You are developing an AI-powered application using Azure Cognitive Services' Computer Vision API. The application processes images to extract textual information using OCR (Optical Character Recognition). After sending an image to the API, the JSON response includes the following fields: 'text', 'boundingBox', and 'confidence'. How should you interpret these fields to provide meaningful insights to the end-user?
- A
'text' contains the extracted textual content from the image.
- B
'boundingBox' specifies the geometric location of the detected text in the image.
- C
'confidence' represents the accuracy percentage of the OCR model for the entire image.
- D
The 'confidence' field can be used to filter out text extractions with low reliability.
- E
'boundingBox' indicates the font style and size of the detected text.
Show answer and explanation
Correct answers: A, B, D
Explanation
When interpreting the Computer Vision API's OCR responses, the 'text' field contains the extracted text, the 'boundingBox' provides spatial coordinates for the detected text, and 'confidence' can be used to assess the reliability of individual text extractions. Understanding these fields helps in providing meaningful insights and enabling accurate text processing in AI applications.
- A. Correct.
'text' contains the textual content extracted from the image. This is correct as it represents the main output of OCR processing.
- B. Correct.
'boundingBox' specifies the geometric location of detected text, including coordinates for the text in the image. This is correct because it helps map the extracted text back to the original image.
- C. Incorrect.
'confidence' does not represent the accuracy percentage for the entire image. Instead, it indicates how confident the model is in detecting and extracting specific text segments.
- D. Correct.
The 'confidence' field can indeed be used to filter out low-reliability text extractions, ensuring higher-quality results for the user.
- E. Incorrect.
'boundingBox' does not provide any information about the font style or size. It only indicates the spatial location of the detected text.