Captions an image by labeling every object the model detects is present within the image.