Optical flow is the apparent motion of brightness between two consecutive frames. For each small patch of the image, the method asks: where did this patch move to in the next frame? The answer is a displacement — a horizontal and a vertical shift — and collecting that answer for every patch produces a dense field of arrows, one per location.
The arrows are not the objects themselves. They are a separate measurement of how the picture is changing. A camera panning across a still wall produces large flow arrows even though nothing in the scene moved; a person walking toward a stationary camera produces arrows radiating outward from the center. This is why flow is described as apparent motion rather than true motion: it captures what the image is doing, which usually but not always matches what the world is doing.
Flow is useful because it strips away appearance and keeps only change. Two very different-looking actions — a bird flapping and a flag waving — can produce similar flow patterns, and two similar-looking scenes can produce very different flow. That complementarity is what makes flow a useful second source of evidence alongside the raw frames.