In object detection output, what is a bounding box?
Show answer and explanation
Answer: D
A bounding box is the geometry that says where an object is. Coordinates may be pixels or normalised fractions of the image, and each box is paired with a class label and a confidence score. Applications use boxes to crop objects, count them, or test whether they overlap a zone. Storage containers, serialisation formats and confidence thresholds are all separate concepts that happen to appear in the same pipeline.
Why the other options are wrong
- A. A container stores files; it does not describe an object's position in an image.
- B. An export format determines how results are serialised, not the coordinates themselves.
- C. The confidence threshold is a decision setting applied to scores, not a geometric region.