In general we don’t have a lot of high-resolution data about things under the ground. The only ways to “see” underground features at any depth are by measuring tiny variations in the Earth’s gravitational field (gravimetry), or by setting off explosive charges and measuring the reflected shockwaves (seismic surveys).
Both of these are mainly done to search for new oil and gas fields. Seismic surveys have better resolution (tens of meters), but cover a relatively small area and are expensive to do. Gravimetry can cover a wide area – it can be done from satellites, though I don’t know if any useful scans of the whole planet have ever been done – but the resolution is even worse, and the data is complicated to interpret.
As I understand it, most places on Earth have never been surveyed in enough detail to detect even large cave systems. Even if the survey was done, it would be proprietary data, so no one would hear about it unless it looked like there was oil there, or if the data were unusual enough that someone could raise quite a lot of money to investigate. And the survey wouldn’t happen in the first place if oil companies didn’t think there was oil there, or if they couldn’t drill there because it was a populated area, for instance.
As you note, caves can be inferred in other ways. In the Mendip hills where I used to live, water is known to circulate underground (scientists can track this with marker isotopes and such), and there are lots of known caves, so it is reasonable to assume there are other, unknown caves. They could be huge and connected, or maybe there are thousands of tiny pockets; there’s no pressing reason for anyone to spend huge amounts to find out. If there weren’t millions of people using water in the area, we wouldn’t even know this much.
One thing I would point out, though, is that if there was weird life in those hidden caves, that might be detected, and attract investigation. Unless the cave biome was completely sealed, and I don’t know if that’s feasible, the water coming out of it would contain spores and metabolites and stuff, so you’d have unusual ecologies around springs at the surface. (But this could still go undetected in a remote area).