A retail chain installs a ceiling-mounted camera at each store entrance. The company wants software that draws a box around every person who walks through the doorway and outputs a running count of visitors per hour, without identifying who any individual shopper is. Which computer vision capability should the company implement?

The correct capability is object detection, because it identifies the location of multiple instances of an object class, such as people, within an image or video frame and can return a bounding box for each one. That is exactly what the retailer needs to locate and tally every visitor passing through the doorway each hour. Facial recognition is incorrect because it is designed to determine or verify a specific person's identity by comparing facial features against a reference, which the retailer explicitly does not want since it only needs a count, not identities. Image classification is incorrect because it assigns one label to an entire image, such as 'busy entrance' or 'empty entrance,' and cannot locate or count multiple individual people within a single frame the way the retailer requires. Optical character recognition is incorrect because it extracts printed or handwritten text from an image, such as reading a sign or a receipt, and has no role in detecting or counting people walking through a doorway. Because the requirement is locating and tallying discrete people in each frame while deliberately avoiding identification, object detection is the only capability that matches both the technical need and the stated privacy constraint.