we were studying utopias / distopias in my english class and we were put into groups for a project. our group chose to play with the idea that justice in a utopian society would be completely impartial, so we tried emulating sentencing guidelines in excel with the idea being that a computer would be the only reliably impartial judge. we started with the code of virginia, throwing in some of our own ideas on how to factor in a person's past history, and ended up with a monumentally complicated spreadsheet. long story short, we worked really hard to try to do a reasonable job and we just weren't able to do it. looking back, there were clearly significant technical flaws in our approach but the lessons were pretty clear: it's a lot harder to automate justice than it looks and the cost of a mistake in data entry could be devastating (at least in our model). our final conclusion was that while a piece of software may be an aid to the sentencing process (automatically looking up minimum and maximum sentences, presenting summaries of similar cases, etc.) there was no way to take a human out of the loop.
i interpreted the response as meaning "i have no idea what you're talking about with regards to allofmp3 being legal", which is a totally valid statement --- nobody has tested the legality of allofmp3 in court. just because you buy something doesn't mean it's legal (e.g: crack cocaine, stolen goods, etc.)
get in touch with our business folks at http://perceptek-robotics.com/ --- my best guess is that this task is about a man week. if they turn you down, i'm open to offering my services as a consultant.
"network video recorders" are becoming pretty standard fair these days. if you pick up a copy of "security technology and design" you'll find it brimming with companies selling various nvr solutions. the plus is that the captured data is all digital and easy for you to use as you see fit. there are several companies which sell solutions built on top of windows (which i'm guessing is what you use). you can make arbitrarily many bitwise copies with zero degradation.
i've never tried doing what you're suggesting under windows. under linux, however, is a different story. all the ingredients you need exist: you'll need a video4linux compliant usb webcam (philips makes a chipset that's in some).
(this captures 1 hour of video using the h263p codec and saves it in a file called video_[4 digit year]_[short month name]_[2 digit month day]_[2 digit hour (24 hour)]_[2 digit minute])
the specifics of dealing with cron are up to you, i'm afraid.
under windows, if you're comfortable using windows scripting to control an application, you might be able to use the playcap app from http://directshownet.sourceforge.net/about.html and control things that way (i.e. the start, stop, file name, etc.)
my experience has been that even tho there's a mild processor hit (20% of a P4 mobile 2GHz) it's a perfectly reasonable hit when you consider that you get far more control over the encoding (i personally favor h264 encoded avis @ about 1mb/s which ends up being far higher quality than an mpeg2 of comparable size).
yes, there are uncrippled machines that can do what you want (and then some). you probably have one sitting on your desk
if you're running under linux, you've got a couple options. kino (http://www.kinodv.org/ will allow you to capture live raw video (plus sound) from a standard dv camera with an ilink (aka 1394a) connection. it takes a little effort to get setup, but it's worth it. you'll then want to use ffmpeg to re-encode the files so that they're less huge and then save the encoded version.
if you have analog cameras, a $50 capture card (we use ati's all-in-wonder) can act as a frame grabber --- it may take a little finagling to get the sound working, but once it's all hooked up you should be good to go. use xawtv to preview and make sure that everything is behaving as expected, then use ffmpeg to capture the video. make sure you encode at fairly high bit rate and be careful about what combinations of codec and containers you choose (in particular, you probably want to stick to msmpeg4v2 encoded.wmv files if you intend the video to be played back on windows machines). if you've installed something like VLC on the playback machines, you can use more interesting codecs like h264 and still achieve quite impressive playback quality at much lower bitrates.
there are ways to do similar things in windows, although i have much less experience doing so and tend to use developers tools (like graphedit) to put together the directshow filters that will capture video and sound from some source, encode, mux, and then output the file. i'm sure that there are pieces of software out there that can do this. if you have access to some it people, writing your own should be fairly easy (there's a handy book on the subject here: http://www.amazon.com/Programming-Microsoft/dp/073 5618216/sr=8-1/qid=1156903037/ref=pd_bbs_1/104-273 5593-2181510?ie=UTF8)
if you're not inclined to build your own solution, virtualdub http://www.virtualdub.org/ may be able to help you. i haven't used it myself, but it's a pretty widely used app.
the one thing to bear in mind with all these proposed solutions is that you're going to want to make sure you've got fairly big and fast disks and quite a lot of space free. you're also going to want to make sure you've got a reliable backup strategy in place since you no longer have the luxury of the original tapes. if you have any other questions, feel free to email me: (my slashdot user name) 'at' yahoo(dot com).
in theory, this is precisely what the cognitive radio (see http://en.wikipedia.org/wiki/Cognitive_radio) would fix. assuming you've got a sufficiently wide spectrum that it can't all be jammed at once, the radio will detect which portions of the spectrum are being jammed and transmit in clear bands. of course, that's all in theory. in practice, who knows?
but why is the bahraini government doing this? are they scared people are going to use google earth to zoom in on paris hilton's naked follies? or are there other reasons, like "national security"?
1) do not build computers for your sister, or any other relative that isn't going to learn to use them appropriately.
2) let dell, or geek squad, or whatever local mom and pop support shop deal with the problem.
i don't think that anyone can fault you for not wanting to deal with this anymore after genuinely making the effort to help your sister and being repaid by ingratitude. once support starts costing your sister, she's going to think twice about using it. she's going to have motivation to become more savvy on her own and you're not going to have to deal with her frustration and ingratitude. sounds like a win all around.
is to have the electors agree to vote in accordance with the popular will of their assigned district. on the one hand, there's an incentive to be somewhat more reasonable about how districts get gerrymandered (a finite supply of partisan voters which would now need to be spread to maximize impact across several districts rather than concentrated in a single district)
that's pretty much my job description too. i've lucked out big time in the company that i work with — the ph.d.s often come to me with hairy questions and i tend to be given the hard theory parts of projects to work. i love it... i'm functioning largely as a ph.d. without the actual letters attached to my name. that said, i'm planning on going back to get the ph.d soon — as i pointed out, DARPA isn't hiring PMs that don't have ph.d.s. ultimately, the jobs that i want to see my self in ("principle investigator" or "director" type roles) are all going to be made a lot easier to achieve with a ph.d. that plus i can "retire" as a professor and live out the rest of my days as a professional dirty old man;)
... a graduate degree is a great thing. It opens doors to jobs which simply would be shut otherwise (e.g: DARPA now only hires Ph.D.s to be program managers) and you can expect a healthy salary premium for those jobs. That said, it takes a particular kind of personality to do well in grad school and to excel at those jobs which require graduate level degrees. If you're in it just for the money, do an MBA, because you are likely to be miserable (and, incidentally, also make the people in your classes miserable) otherwise. Expect to put in 5 to 8 hours of projects and studying in per hour of lecture if you're serious about succeeding. If you aren't comfortable working with theory and concepts at a highly abstract level, you also need to seriously reconsider your motivation for pursuing a graduate degree. If you lack the intellectual curiosity and discipline to seek answers out for yourself, you have no place in grad school. The program that I went through hit the theory hard very early on (mostly as a way of weeding out candidates, the department's philosophy was generally to let most people in and let the core classes separate the wheat from the chaff) and the projects were designed to really emphasize the interface between theory and practice.
In summary: if you're the sort that does well in an R&D environment, then a graduate degree is going to open a lot of doors. Otherwise, you're going to want to steer clear.
take a look inside your passport some time, i expect you'll see a bunch of barcodes (mine does, although it's a south african passport that's about to expire. not sure what the newer passports look like these days).
IXO (DARPA'S Information eXploitation Office) just made awards for their VACE (Video Analysis and Content Extraction) BAA and this sounds a lot like some of the technologies they were trying to develop through that program. I'll have to do more digging, the article itself is somewhat suspect (some jackass with a Ph.D. in *transport systems* flaunting his ignorance of computer vision isn't exactly a good source to quote). I particularly like the bit about "They are limited in their ability to recognize suspicious activities or events." Turns out that he hasn't read Grimson and Stauffer's (fellow MIT alums) papers. Or, you know, about 20-30% of the computer vision literature.
every time i see anything autonomous ground vehicle related, everyone goes off about how fantastic carnegie mellon university is while completely ignoring the fact that sebastian thrun's team at *STANFORD* won GC2. yes, i realize that thrun was red's student for a long time, but when you compare the radically different approaches it's pretty clear that it's thrun's vision, leadership, and team which earned the win. since vw was part of thrun's team, the odds are overwhelmingly in favor that any autonomous control of the 53+1 was largely a product of their collaboration with *STANFORD* and has pretty much bugger all to do with CMU.
pretty much regardless of what you want to do in robotics (other than at the technician level), you're going to need a master's degree and quite possibly a ph.d, depending on where you end up working. part of the process of getting the advanced degree means accepting that you are probably not one of the very few people in the world who are able to master robotics from the high level control / deliberative planning, low level reactive planning, perception, mechanical and electrical control, communications & processing architecture, all the way down to the low level circuits and mechanical systems. so pick something you want to do well in (in my case it's the machine perception and sensing side) and write a master's thesis (or do a master's project) on something relating to that speciality.
regardless of your chosen speciality, you're going to want to get to be really good at math. in particular, linear algebra, calculus (multivariate), trig, analytic geometry, and stats/probabilities (in particular, bayesian thinking). also make sure you are really solid on forward and inverse kinematics. finally, an understanding of communication systems will be incredibly useful.
concurrent with that, see if you can get a job working in the field. drop me an email and i can get you in touch with my company's recruiting/hr people (we qualified 5th at GC2 and we do a lot of UAS an UGV work). of the r&d staff, about 17% hold ph.d.s and over 90% hold at least one master's degree (there are several folks with multiple m.sc. degrees).
finally: good luck! i know i couldn't believe that i'm actually being paid to play with the toys and do cool stuff with them:-)
http://imagemagick.org/ is free, runs on most any platform you'd ever case to use, and implements fuzzy c-means for image segmentation (which basically turns regular photos into "coloring book" versions of themselves). the c-means segmentation may be a bit too coarse for what you want to do. in that case, implementing a mean-shift segmenter (google for comaniciu and meer) will probably do a nice job. hope this helps!
to be fair, the additional hardware capabilities open up a lot of methods which might otherwise be unfeasible for OCR and AI in general. case in point, i worked on a project doing real time detection, tracking, and recognition of highway signs. we had a dual p4 xeon which, at the time, was pretty much the beefiest box we could afford for this project. we would have much preferred to use a mean shift filter to do the edge detection and segmentation in the image because it gave really excellent results. unfortunately, it just wasn't fast enough --- using the mean shift stuff typically slowed the system to 1Hz cycle rate. if we had about twice the computational power, we might have been able to make it work.
this is standard procedure for dvorak --- if you don't believe me, listen to the "this week in tech" (aka twit) podcast. as far as i'm concerned, he gets filed together with used car sales people and 97% of politicians under "people i may or may not swerve to avoid on the street"
this is pretty counter my experience with CCD v. CMOS video cameras. the CMOS cameras that i evaluated about 14 months ago had noticably worse noise characteristics than the CCD cameras. i suppose that it probably doesn't matter too much for a still camera, but for video (at least when i last looked), it wasn't so great.
Slashdot Mirror
we were studying utopias / distopias in my english class and we were put into groups for a project. our group chose to play with the idea that justice in a utopian society would be completely impartial, so we tried emulating sentencing guidelines in excel with the idea being that a computer would be the only reliably impartial judge. we started with the code of virginia, throwing in some of our own ideas on how to factor in a person's past history, and ended up with a monumentally complicated spreadsheet. long story short, we worked really hard to try to do a reasonable job and we just weren't able to do it. looking back, there were clearly significant technical flaws in our approach but the lessons were pretty clear: it's a lot harder to automate justice than it looks and the cost of a mistake in data entry could be devastating (at least in our model). our final conclusion was that while a piece of software may be an aid to the sentencing process (automatically looking up minimum and maximum sentences, presenting summaries of similar cases, etc.) there was no way to take a human out of the loop.
i interpreted the response as meaning "i have no idea what you're talking about with regards to allofmp3 being legal", which is a totally valid statement --- nobody has tested the legality of allofmp3 in court. just because you buy something doesn't mean it's legal (e.g: crack cocaine, stolen goods, etc.)
get in touch with our business folks at http://perceptek-robotics.com/ --- my best guess is that this task is about a man week. if they turn you down, i'm open to offering my services as a consultant.
"network video recorders" are becoming pretty standard fair these days. if you pick up a copy of "security technology and design" you'll find it brimming with companies selling various nvr solutions. the plus is that the captured data is all digital and easy for you to use as you see fit. there are several companies which sell solutions built on top of windows (which i'm guessing is what you use). you can make arbitrarily many bitwise copies with zero degradation.
i've never tried doing what you're suggesting under windows. under linux, however, is a different story. all the ingredients you need exist: you'll need a video4linux compliant usb webcam (philips makes a chipset that's in some).
/dev/correct_video_device -t HH:MM:SS.MS -vcodec YOUR_CODEC your_file.ext
/dev/video0 -t 1:00:00.000 -vcodec h263p -b 1000 video_`date +"%Y_%b_%d_%H_%M"`
:)
to capture for one hour:
ffmpeg -vd
you can put something similar into a cron job to run when you need it, using some clever shell trickery to generate the filenames for you, eg:
ffmpeg -vd
(this captures 1 hour of video using the h263p codec and saves it in a file called video_[4 digit year]_[short month name]_[2 digit month day]_[2 digit hour (24 hour)]_[2 digit minute])
the specifics of dealing with cron are up to you, i'm afraid.
under windows, if you're comfortable using windows scripting to control an application, you might be able to use the playcap app from http://directshownet.sourceforge.net/about.html and control things that way (i.e. the start, stop, file name, etc.)
good luck
my experience has been that even tho there's a mild processor hit (20% of a P4 mobile 2GHz) it's a perfectly reasonable hit when you consider that you get far more control over the encoding (i personally favor h264 encoded avis @ about 1mb/s which ends up being far higher quality than an mpeg2 of comparable size).
yes, there are uncrippled machines that can do what you want (and then some). you probably have one sitting on your desk
.wmv files if you intend the video to be played back on windows machines). if you've installed something like VLC on the playback machines, you can use more interesting codecs like h264 and still achieve quite impressive playback quality at much lower bitrates.
3 5618216/sr=8-1/qid=1156903037/ref=pd_bbs_1/104-273 5593-2181510?ie=UTF8)
if you're running under linux, you've got a couple options. kino (http://www.kinodv.org/ will allow you to capture live raw video (plus sound) from a standard dv camera with an ilink (aka 1394a) connection. it takes a little effort to get setup, but it's worth it. you'll then want to use ffmpeg to re-encode the files so that they're less huge and then save the encoded version.
if you have analog cameras, a $50 capture card (we use ati's all-in-wonder) can act as a frame grabber --- it may take a little finagling to get the sound working, but once it's all hooked up you should be good to go. use xawtv to preview and make sure that everything is behaving as expected, then use ffmpeg to capture the video. make sure you encode at fairly high bit rate and be careful about what combinations of codec and containers you choose (in particular, you probably want to stick to msmpeg4v2 encoded
there are ways to do similar things in windows, although i have much less experience doing so and tend to use developers tools (like graphedit) to put together the directshow filters that will capture video and sound from some source, encode, mux, and then output the file. i'm sure that there are pieces of software out there that can do this. if you have access to some it people, writing your own should be fairly easy (there's a handy book on the subject here: http://www.amazon.com/Programming-Microsoft/dp/07
if you're not inclined to build your own solution, virtualdub http://www.virtualdub.org/ may be able to help you. i haven't used it myself, but it's a pretty widely used app.
the one thing to bear in mind with all these proposed solutions is that you're going to want to make sure you've got fairly big and fast disks and quite a lot of space free. you're also going to want to make sure you've got a reliable backup strategy in place since you no longer have the luxury of the original tapes. if you have any other questions, feel free to email me: (my slashdot user name) 'at' yahoo(dot com).
it might be local if they host a data center in the uk. but then i'm not a lawyer, merely a techy prone to speculation.
i'm partial to solutions involving lots of robots (but then i'm somewhat biased given that my job is teaching robots how to see)
in theory, this is precisely what the cognitive radio (see http://en.wikipedia.org/wiki/Cognitive_radio) would fix. assuming you've got a sufficiently wide spectrum that it can't all be jammed at once, the radio will detect which portions of the spectrum are being jammed and transmit in clear bands. of course, that's all in theory. in practice, who knows?
but why is the bahraini government doing this? are they scared people are going to use google earth to zoom in on paris hilton's naked follies? or are there other reasons, like "national security"?
1) do not build computers for your sister, or any other relative that isn't going to learn to use them appropriately.
2) let dell, or geek squad, or whatever local mom and pop support shop deal with the problem.
i don't think that anyone can fault you for not wanting to deal with this anymore after genuinely making the effort to help your sister and being repaid by ingratitude. once support starts costing your sister, she's going to think twice about using it. she's going to have motivation to become more savvy on her own and you're not going to have to deal with her frustration and ingratitude. sounds like a win all around.
is to have the electors agree to vote in accordance with the popular will of their assigned district. on the one hand, there's an incentive to be somewhat more reasonable about how districts get gerrymandered (a finite supply of partisan voters which would now need to be spread to maximize impact across several districts rather than concentrated in a single district)
that's pretty much my job description too. i've lucked out big time in the company that i work with — the ph.d.s often come to me with hairy questions and i tend to be given the hard theory parts of projects to work. i love it... i'm functioning largely as a ph.d. without the actual letters attached to my name. that said, i'm planning on going back to get the ph.d soon — as i pointed out, DARPA isn't hiring PMs that don't have ph.d.s. ultimately, the jobs that i want to see my self in ("principle investigator" or "director" type roles) are all going to be made a lot easier to achieve with a ph.d. that plus i can "retire" as a professor and live out the rest of my days as a professional dirty old man ;)
... a graduate degree is a great thing. It opens doors to jobs which simply would be shut otherwise (e.g: DARPA now only hires Ph.D.s to be program managers) and you can expect a healthy salary premium for those jobs. That said, it takes a particular kind of personality to do well in grad school and to excel at those jobs which require graduate level degrees. If you're in it just for the money, do an MBA, because you are likely to be miserable (and, incidentally, also make the people in your classes miserable) otherwise. Expect to put in 5 to 8 hours of projects and studying in per hour of lecture if you're serious about succeeding. If you aren't comfortable working with theory and concepts at a highly abstract level, you also need to seriously reconsider your motivation for pursuing a graduate degree. If you lack the intellectual curiosity and discipline to seek answers out for yourself, you have no place in grad school. The program that I went through hit the theory hard very early on (mostly as a way of weeding out candidates, the department's philosophy was generally to let most people in and let the core classes separate the wheat from the chaff) and the projects were designed to really emphasize the interface between theory and practice.
In summary: if you're the sort that does well in an R&D environment, then a graduate degree is going to open a lot of doors. Otherwise, you're going to want to steer clear.
take a look inside your passport some time, i expect you'll see a bunch of barcodes (mine does, although it's a south african passport that's about to expire. not sure what the newer passports look like these days).
IXO (DARPA'S Information eXploitation Office) just made awards for their VACE (Video Analysis and Content Extraction) BAA and this sounds a lot like some of the technologies they were trying to develop through that program. I'll have to do more digging, the article itself is somewhat suspect (some jackass with a Ph.D. in *transport systems* flaunting his ignorance of computer vision isn't exactly a good source to quote). I particularly like the bit about "They are limited in their ability to recognize suspicious activities or events." Turns out that he hasn't read Grimson and Stauffer's (fellow MIT alums) papers. Or, you know, about 20-30% of the computer vision literature.
every time i see anything autonomous ground vehicle related, everyone goes off about how fantastic carnegie mellon university is while completely ignoring the fact that sebastian thrun's team at *STANFORD* won GC2. yes, i realize that thrun was red's student for a long time, but when you compare the radically different approaches it's pretty clear that it's thrun's vision, leadership, and team which earned the win. since vw was part of thrun's team, the odds are overwhelmingly in favor that any autonomous control of the 53+1 was largely a product of their collaboration with *STANFORD* and has pretty much bugger all to do with CMU.
stanford and mit may disagree with you, so would umd and purdue (to lesser degrees tho).
pretty much regardless of what you want to do in robotics (other than at the technician level), you're going to need a master's degree and quite possibly a ph.d, depending on where you end up working. part of the process of getting the advanced degree means accepting that you are probably not one of the very few people in the world who are able to master robotics from the high level control / deliberative planning, low level reactive planning, perception, mechanical and electrical control, communications & processing architecture, all the way down to the low level circuits and mechanical systems. so pick something you want to do well in (in my case it's the machine perception and sensing side) and write a master's thesis (or do a master's project) on something relating to that speciality.
:-)
regardless of your chosen speciality, you're going to want to get to be really good at math. in particular, linear algebra, calculus (multivariate), trig, analytic geometry, and stats/probabilities (in particular, bayesian thinking). also make sure you are really solid on forward and inverse kinematics. finally, an understanding of communication systems will be incredibly useful.
concurrent with that, see if you can get a job working in the field. drop me an email and i can get you in touch with my company's recruiting/hr people (we qualified 5th at GC2 and we do a lot of UAS an UGV work). of the r&d staff, about 17% hold ph.d.s and over 90% hold at least one master's degree (there are several folks with multiple m.sc. degrees).
finally: good luck! i know i couldn't believe that i'm actually being paid to play with the toys and do cool stuff with them
http://imagemagick.org/ is free, runs on most any platform you'd ever case to use, and implements fuzzy c-means for image segmentation (which basically turns regular photos into "coloring book" versions of themselves). the c-means segmentation may be a bit too coarse for what you want to do. in that case, implementing a mean-shift segmenter (google for comaniciu and meer) will probably do a nice job. hope this helps!
... these people never heard about vista ;)
to be fair, the additional hardware capabilities open up a lot of methods which might otherwise be unfeasible for OCR and AI in general. case in point, i worked on a project doing real time detection, tracking, and recognition of highway signs. we had a dual p4 xeon which, at the time, was pretty much the beefiest box we could afford for this project. we would have much preferred to use a mean shift filter to do the edge detection and segmentation in the image because it gave really excellent results. unfortunately, it just wasn't fast enough --- using the mean shift stuff typically slowed the system to 1Hz cycle rate. if we had about twice the computational power, we might have been able to make it work.
this is standard procedure for dvorak --- if you don't believe me, listen to the "this week in tech" (aka twit) podcast. as far as i'm concerned, he gets filed together with used car sales people and 97% of politicians under "people i may or may not swerve to avoid on the street"
this is pretty counter my experience with CCD v. CMOS video cameras. the CMOS cameras that i evaluated about 14 months ago had noticably worse noise characteristics than the CCD cameras. i suppose that it probably doesn't matter too much for a still camera, but for video (at least when i last looked), it wasn't so great.