acts are public already so there is no privacy breach there
Equating an event that happened in public to having a recording of an event is a false equivalence.
An event in public can be witnessed live by people who happen to be there. By contrast, a recording is an artifact that grants its possessor the ability to control to whom, whether or not, and when the event is witnessed. Possessing a recording creates a power imbalance not present when merely witnessing an event. This alone makes having a recording drastically different from merely being a live witness. (There are other differences in degree as well, such as that live witnesses can only collect information if they happen to attend to it, but recordings can be played back repetitively and scrutinized).
the main problem here is the bullying by other people regarding their sexual orientation or even simple things like gastronomic taste which could be protected exactly by the surveillance people who engage bullying
This possibly creates another false equivalence, though it's a bit ambiguous.
If you're arguing that it's okay to bully a homosexual/mock an employee's eating habits, if you could in addition catch the bully/mocker, then I'm curious what sort of ethical system you're employing. I don't see how catching the bully/mocker with the same technology erases the harm to the homosexual/employee. Surely not having harm done in the first place is different, and preferable, at least to the affected party who should for that reason have some say.
If you're arguing that nobody who is recorded would want to bully a homosexual, or mock an employee, then you're living on a different planet from me. The notion that the bully would suppress his behavior while being recorded but the homosexual should be unaffected (i.e., free to show affection) while being recorded is extremely naive. It also misses the point; this is a risk assessment issue. Whereas you're not the one facing the potential harm in this scenario, it's not up to you to assess the risk on behalf of the potentially harmed party.
In the other hand I noticed that a number of people that I know who are strongly against public recording often engage in socially unacceptable behavior like cheating their partners or vandalism.
That observation is meaningless. For whatever reason, a large number of people cheat. If you actually used the fact that a large number of people who are against public recording cheat to conclude something, then you're doing something wrong. To just establish a correlation between cheaters and anti-public recorders, you need to have a comparative analysis across all groups (including pro-public recorders); i.e., you cannot ignore the base rates.
This personal observation lead me to think that people who don’t want to be recorded probably have a guilty mind of something wrong they did or still do in public space and they don’t want this to be known to the public.
Let A be that a person cheats. Let B be that the person is against public recording. Then so far, you had been arguing that A=>B. But here, you're concluding from this that if B, then A. In other words, you are arguing:
A=>B, B, therefore A.
That is affirming the consequent.
As it turns out, the most general concern here isn't merely being exposed for doing something wrong. It's any situation where the revelation of information to some particular party can cause a harm. We're not just talking about people being guilty, we're talking about preventing your friend's abusive ex from learning where your friend lives; keeping despotic anti-butter governments from learning where the butter factories are; keeping mobs of butter-on-the-other-side folk from attacking a butter-on-the-same-side proponent. Keeping a thief from learning mothers' maiden names.
Or perhaps, being exposed to a bully for being homosexual, or to rude coworkers for your culinary habits?