Let's first define "living". A living thing is an object that shares some traits with other living things, e.g. it has a metabolism, it has been "born" (in the broadest sense, including cell division and so on) and it changes throughout it's lifetime and will, eventually, someday die.
Imagine you see a picture of a tiger: You instantly recognize it being a living thing, more specific: Something you call "tiger". You have seen tigers before (or at least heard of them and saw pictures or so). You have somekind of (vague) concept of what a tiger might be and this thing fits percectly in it. And you know that a tiger is a living thing and you believe that what you see is a tiger: So you believe that the thing you see is a living thing called tiger.
Then, a bit later, you see a picture of a Nasobēm. You've never heard of that before (if you did, assume you didn't). You have no concept at all of what a Nasobēm is. You haven't heard anything from it and still you somehow see: It (might be) and it is very like to be a living object. Still, you haven't seen anything resembling something Nasobēm-like ever before in your life.
What exactly are the features we all recognize living things kind of acurate (even though when they don't exist in reality and are, just like the Nasobēm, just imaginary animals someone invented)?
Of course, these are not perfect algorithms, but just kind-of-okay-heuristics, meaning, their result might not be perfect, but it's good enough for "everyday-usage". But still: It has to rely on things that we ourselves observed and considered "living", but what phenomenons do we actually use? Remember, this is just a still-picture, you cannot see anything moving (which might be an indicator for living things in one way or another).