That's essentially a question for Laurent, why do so function have an overload for (float, float) while others have just the (sf::Vector2f)? (I bet there have been a few posts about this already...)
Given that I haven't watched the video and thus don't know if he has actually said something about it, I simply assume, that he used the uniform initialization (curly brackets) for the setSize function, because it only accepts an sf::Vector2f, which will get created on the fly with those brackets and you don't have to type the full class name. For the setOrigin function he uses the "native" overload of (float, float), because it's there.
Some might argue about not having to construct a temporary sf::Vector2f object, but that really doesn't matter. Much more interesting would be a discussion about consistency, like when should one use the overloads (if existing) and when should one use the vector directly.
In any case the width and height, could've also been packaged into a sf::Vector2f in the first place, making the code more compact, but maybe introducing more challenge for teaching the code.