In the literature, I see "corner detector" and "feature detector" are interchangeably used for these methods e.g. Harris, Shi-Tomasi, FAST, etc.
What is the difference between a corner detector and a feature detector? Are corner detectors (Harris, Shi-Tomasi, etc.) one kind of feature detector?
A couple of confusions here.
Features refer to some form of lower dimensional descriptors that explain a (potentially local) region of interest. They are useful in converting the appearance into certain signatures that are easier to handle / more robust than naively using pixels.
What you are trying to ask is probably keypoint detection. This term is used to refer to certain points of interest in the image, that are interesting or worthy to extract features at. Indeed, this is a more general term for corner detection. Most of the time (especially nowadays in the context of deep learning), keypoints need not be some geometrically salient corners but rather points that are critical in describing the image/signal as a whole. The state of the art methods learn this notion of saliency and critical-ity.
In fact, the term feature detection should probably refer to the combination of two steps: first detecting the keypoints and subsequently extracting descriptors on those locations.
Hope this makes sense.
