How to study computer vision and machine learning?

Question

OpenCV

Hi all!
I am sincerely interested in learning and solving problems using computer vision and machine learning (deep learning in particular)
I repeated many examples from the Internet, trained my own non-complex neural networks, or already existing models, for test problems and for the sake of interest, explored this area, easily accessible to everyone, methods However, when trying to solve problems that are close to working, I encounter insurmountable difficulties, and the usual googling
does not work.
, however, the quality of work did not satisfy me at all (the quality of recognition was very dependent on lighting, etc.), I wanted to do better
So now I also have a question about contour recognition using opencv

. The main question is what to do to be able to solve almost any problem of this kind? In order to understand in what cases what kind of neural network is needed, and how to make it, in what cases and what transformations and methods should be used, so as not to beat my head against the wall, but to be more or less able to solve the tasks that are facing me?

Answer 1

Your context and background is not very clear. I dare to suggest that in your case, the best way would be to find yourself a mentor. He will individually prompt, direct and save a lot of time. And by the way, it will not necessarily be expensive (and maybe even free).
As an alternative way - find a strong team and go to work there, just a jun for food. In half a year or a year of diligent study, you can get involved and absorb the necessary knowledge and the level from which you can independently build on when making decisions.

Answer 2

Finish CS center https://compscicenter.ru/

Answer 3

Read more theory from (sorry for the tautology) pattern recognition theory?
Search for materials on specialized resources?
Spend more time tweaking and experimenting with different approaches?
Maybe try everything at once?
There is no magic pill, everything comes through experience, study and work + a lot of time devoted to one topic or even one task. This is the only way to achieve something.