I think one of the prerequisites to be able to learn better or faster is to solve internal and external sensory interferences.
So figure your senses. Self accomodate that.
And whatever internal ails you have, get something to alleviate that.
Afterwards, you had to know how much attention you can sustain or regulate and your working memory -- before you can observe, learn certain cues and start practicing.
Else, you will have a harder time if you don't know what interferes or drains you.
You'll be too slow, too overwhelmed, too inattentive, too tunneled in, too distracted. No room to process even if you have prior experiences.
Afterwards the mindset around this involves being both egocentric and allocentric, simultaneously or switching between the two rapidly.
But for the sake of learning others, figure out how to be a bit allocentric. Get out of your head so to speak.
Or better yet -- get rid of whatever hung ups you have making your head prioritizes over different things so you'll be less distracted.
I have different priorities in socializing.
I don't read or watch about it, only that I just start local no matter how much my head fancies itself it's the west -- I'm not in the west.
I'm mainly a naturally passive actor (because of my asocial nature) unless the situation demands certain urgency to initiate, or if I'm bored enough to be.
I can see flirting. I just don't care.
And I let anyone know that I will say no.
Ultimately, I'm an autonomous agent who can choose when and when not to play along with social creatures.
I would not recommend my own ways on how I started to figure the processes and how I managed to learn body language.
... Unless jaywalking is not illegal and not deadly from where you came from.
Really, I was learning from a very simple way of learning social cues naturally, knowing how many senses I need, along with the subtleties and the reaction time needed to predict or know the message, and very much not get myself killed or cause an accident, and etc.
While being overwhelmed, fatigued, almost partially shutting down, etc.
It was so simple enough, that it's actually small enough to have a room for it in a middle of all the internal and external storms -- save for a full-blown meltdown where self-preservation and self control is thrown out of the window.
Then compared that type of human interaction to semi-long term, closer in proximity, more complex mind reading, more parts; face, hands, arms, etc.. , open-ended and multiple involvement interactions that 'mattered more', and layers and layers of contexts --
Than near split second interaction that may only involved movement and vehicle sounds with a clear end goal and possibly not seeing some driver's face and their vehicle ever again.
I don't know the substitute for that, however, other than possibly being a driver...
And use that as some sort to base to work one's way up to more complex interactions.