The first reviewing phase for the International Conference on Computer Vision (ICCV) ended this weekend. This first page was intended to cull those papers that clearly didn't fit the ICCV template without even providing them with a full normal review. ICCV should accept a couple of hundred papers, but there were about 1200 submission, so culling some papers quickly is attractive. The downside is that some of these culled authors probably worked hard on their papers, but I appears that they won't get back complete reviews.

The 2007 Computer and Robot Vision (CRV) just finished. It is an annual conference sponsored by the Canadian Image Processing and Pattern Recognition Society (CIPPRS, which I happen to be the president of, so be aware this is not objective). CRV took place in Montreal combined with AI 2007, Graphics Interface and the Precarn conference. From all accounts the conference combination worked well and there is a desire to hold all four conferences together again next year where the venue will be Windsor, Ontario. Having Precarn in the mix had the conference overall a bit more professional and upscale, and people were nervous about possible increased costs, but it all worked out nicely. Windsor is certain to be the venue for the next AI/GI/CRV, but Precarn still needs to decide if they will join.

Speakers included Michael Black who spoke about Markov Random Fields (MRF's) for image restoration. In addition to showing of some of his own ideas Michael provided a nice overview of the area and discussed methods to avoid explicit computation of the partition function. This material was especially appealing to me since it is related to the research of my former student Abril Torres (Luz-Abril Torres-Mendez) who worked with me on the recovery and restoration of range images using MRF's. Invited speakers included Martial Hebert and Larry Matthies from CMU and JPL respectively, who each also gave very nice talks on image motion analysis and on vision based navigation and scene understanding. I sadly missed most of Larry's talk due to an administrative meeting, but the people who talk to me about it were very enthusiastic, which is no surprise since Mars landing applications are always good and Larry delivers a consistenly great presentation. Martial actually gave 2 different long talks, both of which included a nice restrospective introduction as well as some cool very recent results (some of which were only archieved the week of the conference and have not yet appeared in print).

The one minor downer was an expensive banquet at the Marriott that featured a really absymally tough dry capon (chicken). I rarely focus on food, but this thing was truly memorable.

Other features of the conference were a trade show and an open house at the McGill Research Center for Intelligent Machines (CIM). Both had pretty good attendance and seemed to go over quite well.

(Read more about Abril's work on MRF's here.)

The Convention on International Trade in Endangered Species (CITES) will not be adding red and pink coral to the list of protected species this year. The June 2007 CITES meeting just concluded and despite a proposal to limit the trde in red, pink and other corals which seemed destined to pass, the proposal was overturned at the last minute.

The is more bad news for corals. Reefs are already under immense pressure and the bulk of coral reefs that were around when I was a teenager are either destroyed, being severly pressured or under imminent threat.

Red does not, in fact, actually grow on typical familar reefs. It grows in deep water, and grows very slowly. Because it is rare, there are all kinds of ridiculous stories about how it can be used for everything from medecine to romance, but most is used for ornamental jewelery. Supposed red coral nacklaces are offered for sale very widely and in many countries. In case you are tempted to buy it, note that most red coral that is offered for sale is fake and overpriced, not passing it up is both environmentally and economically sound.

Incidentally, a bid to add protection for the Black Rino also was not successful (see the Zambezi society for a horrible rhino poaching photo and background on the UK Adopt-a-Scout ranger support program where you can help). Some good news: the Japanese and Icelandic governments did not succeed in another sleazy and disgusting attempt to remove whale protections. Thank goodness.

CITES is the biggest international trade agreement, but not the only one. Intenational transport of CITES restricted species, or products of such species, (e.g ivory) is a federal offense in many countries (such as Canada and the USA). Species that are not on the CITES list (such as red coral) may still have restricted trade regulations applied by specific countries.

CBC, the Canadian Broadcasting Company, has a feature on consumer robots on their on-line edition today. It includes articles discussing the destiny and state of consumer robotics, as well as the need for investment in robotics research and the robotics industry to stay in competition.

It also has a nice bit on robot lore (i.e. notable robots from
fiction) as well as a popular robotics quiz. There photo gallery on robots from "fiction to fact" is a bit of a disappointment though: is has a lot of standard fictional movie robots, but misses several of the more exotic yet interesting ones (for example the robots in the movie "Silent Running" or Gort). It also has very limited coverage of real robots, missing all interesting research robots and all interesting industrial and military robots, which is quite a set of omissions.

A very surprising meta-analysis was reported in the journal Nature. Apparently a group at Bristol University and Cardiff University, UK found that cannabis use was linked to much higher incidence of psychotic illness in later life. Apparently even the most minimal marijuana use increases the chance of psychosis by 41%. This is a meta-analysis that combines the results from 35 other studies, and which controls for intelligence, substance abuse of other kinds, and other factors. The result is particularly surprising since it is very much at odds with previous work, including a landmark study in the Lancet several years ago, but it looks at more extensive and perhaps more subtle data.

This does not show that there is necessarily a cause-and-effect relationship, just that there is a statistical connection. An established psychologist (my mother) suggested to me that the explanation is that people who are tending towards crazy will try just about anything, while those are are not cannot be predicted, which pretty much explains this correlation. [She really is a professor emeritus of psychology at the Universite de Montreal.]

I should note it has no personal bearing on any lifestyle choice I need to consider for myself; I am mentioning it only because of its broad significance.

References:

http://www.nature.com/news/2007/070723/full/070723-11.html [Nature.com]

The Lancet podcast on this work at http://podcast.thelancet.com/audio/lancet/2007/9584_28july.mp3 [ thelancet.com]

The slate of invited keynote speakers for RSS 2008 (Robotics Science and Systems) is starting to get populated. One of the speakers will be Kevin O'Regan. He does very influential work on human perception, specifically including visual attention, color perception, and change blindness. Change blindness deals with our use of attention and the fact that if our attention is misdirected in the right way we can fail to observe truly significant changes in the visual world.

There are a slew of other very exciting and diverse speakers also planned for RSS. One of the strategies for RSS keynote speakers is to invite well-established figures who are "tangential" to the conventional robotics community, and who can introduce new ideas, issues and research.