The position you describe is a non-naturalist one defined most succinctly by G.E Moore in what has become known as the "Open Question Argument". Essentially, Moore argued that when we say something is 'good' we have not defined 'good' other than in a tautological sense. For example, if we say personal well-being is good, and declare this to be a fact, then personal well-being must become synonymous with good, so the statement above would read "good is good". This would obviously be uninformative (a closed question), whereas, Moore argues, "personal well-being is good" is a statement about which one might debate (an open question) and therefore has not defined 'good' analytically.
This approach, however, is not without it's flaws. Bertrand Russell argues in his Elements of Ethics, that 'Good' can be defined by examples designed to generate the idea of good in our minds in much the same way as we cannot 'define' the colour red other than by showing someone a collection of things all of which are red. By this notion, it doesn't matter that we cannot linguistically express 'good' other than by tautology because all the sentences we use are doing is helping to bring the concept to mind.
It can also be refuted by, at some point on the open question, closing it empirically, or by some belief statement. If, for example, one were to ask whether Queen Elizabeth were the Queen of England in 2017, that might be a meaningful question despite the fact that some empirical investigation might yield the knowledge that the two concepts "Queen Elizabeth" and "The Queen of England in 2017" were, in fact synonymous.
Finally, you could resolve the issue in materialism. You say 'Good' and 'Bad' might be "...just our perception of an action" as if it were possible, or perhaps desirable, that they were something more than that, but we are all human beings, and it is reasonable to suppose that the concept of 'Goodness' evolved through natural selection to serve some evolutionary purpose and so whilst it may be "just" our own personal opinion, that opinion is not only likely to be very similar among all humans, but more importantly, we could say with some authority that a person claiming x is good, where x is something most people consider bad, is likely to be deluding themselves, rather than making an honest statement about their internal state of mind.
It may well be that 'Good' and 'Bad' are just our own personal opinions, but that may still leave them as definitions, described by example, which one could use to make objective moral judgements.