My first papre was publshed

Oh my god. I just read the abstract of the first journal paper I have ever published (not counting my writings for the University's weekly, the Delta). It is horrible writing! After a long and tedious process of writing, revising, co-editing, and revising some more it is unreadable! The present and past tense are alternated seemingly at random. Bizarre sentences arose from disputes and compromises. Strange ambiguities must leave even the most forgiving reader puzzled and irritated. I promise here and now to do a better job next time. Well, leaving the abstract aside, the paper contains the results of my first experiments, which are not entirely uninteresting I think, regarding the perception of signs and fidgeting. It is called 'When and how well do people see the onset of gestures'. To explain what it is about let me give you a little task. Next time when you are away from your computer and among the people, look for two or three people grouped together, talking or otherwise. Preferably they are at enough distance so as not to be overheard (or take offense at your snooping). Now, watch their movements, their hands. And as soon as you see a gesture you knock on something. If you do this you will probably experience three things. First, it is an easy task. Second, you will probably ignore the fidgeting. Third, if you ask a friend to join your little game of gesture-spotting (as I have done on many occasions), he will probably knock in response to the same movements that you do. Now, the ability to spot gestures, and ignore fidgeting, may seem trivial to you, but imagine you were a robot. Do you think computer vision experts would have been able to program you with this ability? At the moment, they don't even have a clue how to solve this puzzle. You would be forced to analyze any movement as 'possibly a gesture I might have to respond to'. And perhaps you would wish you had something to filter out the 50% 'just fidgeting' (estimate by intuition) to save you time that you need to keep from bumping into lantern posts and such.

Jelgerhuis and the Eloquence of the Body

Here is a wonderful video that sums up and illustrates the main ideas about gestures that are available in the work of Johannes Jelgerhuis, a famous artist of this time as a painter, illustrator and actor. At the same time the video follows the story of a recent book by Herman Roodenburg, Eloquence of the Body, which is however mentioned nowhere. But the illustrations are all in that book and the ideas are the same as well, although I am only halfway in the book so far. I scanned the book written by Jelgerhuis between 1827 and 1830, of which an original copy is available in the library of Utrecht. It is a wonderful series of ´lessons´ about gesture, posture and mimicry with a large collection of plates, witch Jelgerhuis etched himself. There is also a 1970 reprint avaliable with Dutch and English summaries and all the plates. Furthermore there is a full translation, with annotations, available in Golding's book on classicistic acting. The ideas of the eighteenth century about the 'welstand' in Holland, or more generally about appropriate bodily behavior for the upper classes are inspiring. They contain an appealing myth: acquire a certain second nature of standing, walking, dancing, etc. and you will be lifted in society. People will value your behavior and may consider you their peers. Yet at the same time, paradoxically, such a second nature was only attainable for those of the right first nature. If you didn't have the right breeding then you might as well forget it. Ever since I read the books I have been trying to walk more upright, keep my feet at right angles, and behave and talk as elegant as I can. Alas, I slip into my burly uncivilized manners as often as not. I blame my parents for not putting me into a straightjacket as a boy and sending me off to dancing and fencing lessons (I forgive them for my genes, they were all they had to give). But Jelgerhuis' book is not about life. It is about acting. He wrote about the art of gesture on stage (see this earlier entry for links to Johann Jakob Engel, but also to the memory of Dene Barnett). But then, life and acting are not exactly two mutually exclusive categories. Literature JELGERHUIS RZ, JOHANNES Theoretische lessen over de gesticulatie en mimiek, gegeven aan de kweekelingen van het fonds ter opleiding en onderrigting van tooneel-kunstenaars aan den stads schouwburg te Amsterdam. Amsterdam, P. Meijer Warnars, [1827-1830]. Roodenburg, H.W. (2004) The Eloquence of the Body. Studies on Gesture in the Dutch Republic. Waanders, Zwolle, 208 p. 1e dr. (Studies in Netherlandish Art and Cultural History, nr. 6) Golding, Alfred Siemon; Jelgerhuis, J. (1984). Classicistic Acting: Two Centuries of a Performance Tradition at the Amsterdam Schouwburg to Which Is Appended an Annotated Translation of the Lessons on the Principles of Gesticulation and Mimic Expression of Johannes Jelgerhuis, Rz. University Press of America

Wii-Gesture Control for Robots

Here is a guy called roschler demonstrating how to (learn to) control an i-Sobot, the world's smallest humanoid robot, with Wii-gestures instead of using a complicated remote control. Because the robot's routines are mostly gestures you can create commands for them through imitation or iconicity. Probably, actions will often have to be simplified, since I do not expect people will want to make an actual somersault to tell the robot to do that. Also, for certain actions or scripts I imagine that the gestures will become arbitrary and not really intuitive, but probably still easier to use than the alternative RC. Robot control, another good niche for gesture recognition? i-SOBOT by ThinkGeek: World's Smallest Fully Articulated Humanoid Robot Buy an i-SOBOT at Amazon for about $250 Watch a fight: i-SOBOT vs. Godzilla More about Robodance, a.k.a. Robosapiens Dance Machine, which can also be used to control WowWee Robots.

ZCam 3D Gesture Recognition

The ZCam, from 3DV Systems, is featuring in several recent movie demos on the YouTube. It is a camera with onboard technology (it emits infrared pulses and catches refections) that not only provides RGB values for a matrix of pixels, but also a Z-score, i.e. the depth of that pixel. It is reported to be quite accurate. It is also reported to become cheaper, and therefore more widely available to, for example, game developers. The ZCam, small but deep? So far, there are demos shown of a squash game, a boxing game, and a flight simulator. But the website from the company has a more extensive video gallery. Here is a nice review on Gizmodo, where they always keep an eye out for interesting gadgets like this. Crowd response has been positive so far but it remains to be seen whether this technology will be adopted in the market. It is at the moment a bit hard to say how well the technology works, what sort of drawbacks there are, etc. But if it does work well than I think it has less disadvantages than stereo-cameras. I have some experience with a stereo-camera (two synchronized cameras with a software algorithm to integrate their output and obtain depth). This is expensive, costs processing power, costs time to configure, is unreliable (it should not be moved or even touched after configuration), and it relies on skin color segmentation to track faces and hands. But what about the ZCam, what are users required to do or not to do, that is the question.

Emblems in action

See here a nice video on the YouTube featuring a lot of gestures (it is a group of Sicilians at the dinner table). Thanks to Kensy Cooperrider and his embodied talk weblog for digging up some video gems. A Sicilian Argument One of the things I like in this video is that you can see some emblems in action, such as the mano bursa (grip hand). It is made near the end of the video by the man in the blue shirt. Kendon (1995, 2004) wrote extensively on how gestures like the ones you see here can be used with a high degree of conventionality. The ring hand alone is used in a variety of conventional ways. It is not hard to imagine that indeed this group of people knows how to make their gestures count.

Name calling in SLN

There is an interesting initiative to include name calling or swear signs in the SLN lessons for parents of Deaf kids was organised by a group of parents of Deaf kids. According to their leaflet (in Dutch) they will be treating the following signs in the first lesson: klootzak (dysphemism for scrotum), idioot (idiot), and sukkel (idiot/loser). See posts on Doof.nl and FOK (with many comments). It made me wonder how much of the vulgar Dutch emblems are part of the SLN lexicon. In fact, I am quite curious in general about the relation between signs and a host culture's emblems (which according to Ekman & Friesen (1969) and Morris et al. (1979) are a recognizable set of symbolic gestures with a conventionalized meaning in a culture that can be used to substitute words or parts of speech. Kendon (1984, 1992) calls them 'quotable gestures'). I will check the set of 20 key gestures from Morris et al. (1979) against an SLN dictionary (Standaard Lexicon NGT, deel 1+2, NGC, 2002) to see which of those 20 emblems are also SLN signs. It should be noted that according to Morris et al. only 12 of the gestures are commonly known in the Netherlands, which I checked with 'Het Gebarenboekje' by Andrea & De Boer (1993) to see if they concurred, which by and large they did. One gesture is at least a well known SLN sign: the Fig. Do you know what it means in the Netherlands and in SLN? (Source: Morris et al. 1979)

Medal of Honor Heroes 2 - Gesture Trailer

There is a new release in the game series Medal of Honor (see also Medal of Honor Vanguard), which again bets heavily on the Wii gesture control.

Trailer: Experience WWII with an all new Wii-mote control scheme!

GameTrailers commenter: For once a Wii control scheme that actually looks intuitive and intricate as opposed to gimmicky. Ill be picking this up tomorrow

Lehmann's Perfect Finger Camouflage

Here is one of the nicest 'Fingers in camouflage' I have ever seen on video.

Lehmann middle finger
Uploaded by goalbumbum

Offside Bundesliga: When confronted Lehmann said: “It’s nonsense. I’ve never done anything like that”. It supposedly happened after one of Lehmann’s risky trademark excursions outside of the box, during the Germany - Cyprus match on Saturday (4-0). The crowd in Hannover started Robert Enke chants. Enke happens to be Hannover’s goalkeeper and Jogi Löw’s current third choice behind Lehmann and Hildebrand.

Lehmann, whose workout was also a bit ambiguous for some, has shown himself a true master of the camouflaged gesture. Playing into people's increased sensitivity he manages to insult those he wants to insult, while he can claim innocence in public. Bravo, a perfect grasp of the perception of insults. Much better than Joe Nedney, who got a $7.500 fine (which is average, see these other fines for similar offences) for his lousy camouflage attempt: Clearly the man is scratching his head (source) The only examples I have of an even more subtle camouflage is when a man painted an abstract work of art, a huge cactus, on the side of his house facing his (complaining) neighbors. They were highly offended, but he got away with it (and eventually removed it).

The Acceptability of Sign Manipulations

My latest research revolved around the question 'when is a sign production still acceptable', or rather, given that context and application of different norms have a big influence on acceptability, 'which (types of) variations are more acceptable?' We ran an experiment which I think has yielded interesting results. Tuesday Jan 22 I will be giving a presentation at the MPI Nijmegen about the experiment and the results as part of the Nijmegen Gesture Centre Lecture Series 2007. This will be in English. At our own workshop 'Een mooi gebaar 2007' I also gave preliminary results in a short 10 minute presentation in Dutch.

Gestures are Sterile, Touching is not

At the renowned Fraunhofer institute they may have built a killer gesture app: Gesture control that lets surgeons control a 3D-display of a head (for example) during surgery while remaining sterile (touching buttons would break sterility, I guess). Rotate the 3D image by gesturing (source)

Press Release: Non-contact image control As if by magic, the three-dimensional CAT scan image rotates before the physician’s eyes – merely by pointing a finger. This form of non-contact control is ideal in an operating room, where it can deliver useful information without compromising the sterile work environment. The physician leans back in a chair and studies the three-dimensional image floating before his eyes. After a little reflection, he raises a finger and points at a virtual button, likewise floating in the air. At the physician’s command, the CAT scan image rotates from right to left or up and down – precisely following the movement of his finger. In this way, he can easily detect any irregularities in the tissue structure. With another gesture, he can click on to the next image. Later, in the operating room, the surgeon can continue to refer to the scanner images. Using gesture control to rotate the images, he can look at the scan of the patient’s organs from the same perspective as he sees them on the operating table. There is no risk of contaminating his sterile gloves, because there is no mouse or keyboard involved. But how does the system know which way the finger is pointing? “There are two cameras installed above the display that projects the three-dimensional image,” explains Wolfgang Schlaak, who heads the department at the Fraunhofer Institute for Telecommunications, Heinrich-Hertz-Institut HHI in Berlin that developed the display. “Since each camera sees the pointing finger from a different angle, image processing software can then identify its exact position in space.” The cameras record one hundred frames per minute. A third camera, integrated in the frame of the display, scans the user’s face and eyes at the same frequency. The associated software immediately identifies the inclination of the person’s head and the direction in which the eyes are focused, and generates the appropriate pair of stereoscopic images, one for the left eye and one for the right. If the person moves their head a couple of inches to the side, the system instantly adapts the images. “In this way, the user always sees a high-quality three-dimensional image on the display, even while moving about. This is essential in an operating theater, and allows the physician to act naturally when carrying out routine tasks,” says Schlaak. “The unique feature of this system is that it combines a 3-D display screen with a non-contact user interface.” The three-dimensional display costs significantly less than conventional 3-D screens of comparable quality. Schlaak is convinced that “this makes our gesture-controlled 3-D display an affordable option even for smaller medical practices.” The research team will be presenting its prototype at the MEDICA trade fair from November 14 to 17, 2007, in Düsseldorf (Hall 16, Stand D55). Schlaak hopes to be able to commercialize the system within a year or so.

Things like this are probably the best bet for the near future of gesture recognition. Niche applications that exploit some specific benefit of using gestures instead of (or besides) other, more mundane interface technology. The biggest hit in gesture land is without a doubt the Nintendo Wii, which exploits another unique selling point of gestures: a higher (or more representative) physical involvement leading to a better 'experience' of a game. It specifically targets gamers who are interested in fun and exercise in a social context. I doubt that hardcore gamers, intent on getting to higher levels of killing sprees, will be very keen on the Wii. And so it will probably remain for the near future. Like with speech recognition, gesture recognition will have to find some nice niches to live in and multiply. Maybe one day, the general conditions will change (ubiquitous camera viewpoints? intention-aware machines?) and gesture can become the dominant form of HCI, driving buttons to niche applications. I wouldn't bet on it right now, though.