A photo storage app scans each uploaded picture and assigns every tag that applies. A single photo from a birthday party held at the beach might receive the tags 'outdoor,' 'people,' 'cake,' and 'water' simultaneously, rather than being forced into just one category. Which capability produces this kind of output?

The correct capability is multi-label image classification—assigning more than one applicable tag to a single image. The scenario describes a model that assigns several independent tags to the same photo at once, none of which exclude the others, which is the definition of multi-label image classification: the whole image gets multiple category labels simultaneously rather than being forced into a single mutually exclusive class. Optical character recognition—extracting printed or handwritten text visible in the image—is incorrect because it extracts text characters that appear in an image, such as a sign or label, and would not produce conceptual tags like 'outdoor' or 'cake' unless those words were literally printed somewhere in the photo. Semantic segmentation—labeling every individual pixel in the image with a category—is incorrect because it assigns a category to every individual pixel, producing a detailed map of where each region is, rather than a short list of whole-image tags. Object detection—drawing a bounding box around each distinct item in the image—is incorrect because it locates specific items with bounding boxes and identifies where each one sits in the frame, which is a different output than a flat set of descriptive tags applied to the picture as a whole with no location information. Only multi-label classification matches assigning several non-exclusive tags to one entire image.