tony schultz at the the winger blog writes about computer vision dance graphs:
My research uses computer vision algorithms to represent the body as a set of chromatic particles. Once the body is reduced to numbers it becomes possible to automatically recognize different poses. Once these landmarks are identified the computer can generate a map of the movement space in the form of a dance graph. Watch an example here.
computer vision is an umbrella term for image processing, image analysis and machine vision. i have become familiar with these terms over the last few year by chatting with the programmers who have worked on the same projects as me. looking at the sample tony seems to be using feature detection / blob extraction you can see where it doesn’t work properly, and gets confused (e.g. the legs become one particle).
as a dancer i would find such as system of notation hard to use, it does not give me enough information about how the body is moving. the flat particles are hard to read and there is no spatial reference. the only reason i vaguely understand it is because i am familiar with the style of choreography. i’m really not sure how useful this type software really is.
writing choreography and the body isn’t as automatic as tony suggests, there is much more to the dancing body than numerical analysis can graph.
diverse variations 
touche anais. i am happy to see you crit my work. though it may not be useful for you as a user it seems to be useful to you as a blogger. i am flattered you used it for your first real post.
in defense of my work let me just say that my mathematical model is just that, a model. it no more real than a spinor is a an representation of an electron. the model does not get “confused” or fail to “work properly” when the legs fuse together, it simply represents the legs as a single particle when there is occlusion between mirror limbs. this single particle representation of the legs together is sufficient for the detection algorithm to function and for the viewer to perceive what is happening, and thats all the only work i need to get done.
i am not sure what you mean when you refer to my style of choreography. the video presents only a single path derived from the graph extracted from a set of improvisational data. my other posts on “the winger” clarify and give clear examples of this harvesting technique and what i mean by a dance graph. indeed there is “much more to the dancing body than numerical analysis can graph” though this, and my related dance graphing tools, have peaked the interest of several accomplished choreographers including sara rudner and twyla tharp. i am finding many choreographers think of their compositional landscape in terms of these types of markov networks, though they might not call them that.
i look forward to continued discourse with you. it would be great to see some footage of your dancing on your blog. drop a comment on “the winger” next time you have a question about my work. i am always glad to have the attention. 😉