Slashdot Mirror

← Back to Domains

Domain: tjs.org

Stories and comments across the archive that link to tjs.org.

Comments · 12

  1. Alternate mirror (better bandwidth) by moyix · · Score: 3, Informative · on Linux at Industrial Light and Magic

    Since my DSL upload (a puny 15 k/s) is flooded right now, have a shot at my school's fractional T1 (60 k/s).

    Here.

  2. Re:I like the daily log of cat activity by moyix · · Score: 1 · on Cat Recognition Algorithms?

    In my experience, false rejections are a LOT more common than false acceptances when doing any type of automatic image matching (face recognition, cat recognition, whatever). On the system I set up (a really hackish and amateur one, to be sure), we only got one false acceptance and a pretty large number of false rejections.

    If you want to know more about what I was working on, um... too bad, I haven't documented it except in this slashdot comment.

    The link to the source code given in that comment is outdated, though, use this one.

  3. Re:the algorithm... by moyix · · Score: 1 · on Cat Recognition Algorithms?

    I do have some experience with this, as a matter of fact. A friend and I, armed with an aging color quickcam and a bit of ingenuity, set out to build just such a home brew recognition system.

    We started out by going to the VisMod page at MIT (too lazy to link at the moment), where we found the software package "facerec", which claimed to do simple face recognition using the Eigenface method (there are plenty of papers on this in PS and PDF format at the VisMod page). After a bunch of tweaking to get the software (originially designed to work with 10 year old UNIX machines) to work with Linux/BSD, and figuring out what their image format was, we did a bit of testing.

    The system worked pretty well, but we found that the major flaw was that it didn't do any "normalization" of the images; that is, it directly compared them without first checking to see if they were in the right position or even if there was anyone in the picture at all!

    Right now, we're looking into face detection, which should allow us to crop the picture to include only the face, and then scale it to match the database images, before actually doing Eigenface comparisons.

    The field is a fascinating one, especially when you consider the recent privacy concerns. An interesting tidbit: the very same company that was monitoring people at the SuperBowl uses (in modified form) software licensed from MIT--in other words, they were working with roughly the same base we were.

    You can grab facerec, updated to work with Linux and with a bunch of bugfixes (though sadly, the X display still doesn't work--any fixes to that would be appreciated) here

  4. Some work done by moyix · · Score: 1 · on Linux and Biometrics?

    I've been doing a little bit of work in this area--it seems that the VisMod group at MIT released a simple face recognition system about 10 years ago. It uses the eigenface method, which is generally considered the standard (though there have been other models proposed).

    I've ported the code to Linux and BSD, and it works all right, though it takes a lot of configuring, and uses a strange format for the images. I'll get around to making a HOWTO at some point, but until then, if anyone wants to play around with the program, they can grab the source for Linux and BSD.

    Since these are open source, and the method is fairly easy to understand (there are docs all around about it), I hope that some OSS programmers will take some time to improve on this.

  5. Some work done by moyix · · Score: 1 · on Linux and Biometrics?

    I've been doing a little bit of work in this area--it seems that the VisMod group at MIT released a simple face recognition system about 10 years ago. It uses the eigenface method, which is generally considered the standard (though there have been other models proposed).

    I've ported the code to Linux and BSD, and it works all right, though it takes a lot of configuring, and uses a strange format for the images. I'll get around to making a HOWTO at some point, but until then, if anyone wants to play around with the program, they can grab the source for Linux and BSD.

    Since these are open source, and the method is fairly easy to understand (there are docs all around about it), I hope that some OSS programmers will take some time to improve on this.

  6. Don't see the need by moyix · · Score: 1 · on Linux on Older Hardware

    I've installed debian personally on a wide variety of low end machines:

    • Our main school server, colby.tjs.org, a P133, 64M of RAM, newer 20G HD. This runs DNS, POP3, SSH, Telnet, and Web without a hiccup. We came across a cheap IDE RAID card, and we're adding that in now.
    • A Pentium 75 which recently died from flaky hardware.
    • A Pentium 75 with all the spare SIMMS we could cram into it.
    • A 486, 14M of RAM, 275M HDD. It web serves.
    • A 386, specs forgotten since I accidentally killed it (oops!)

    In fact, the computers I work with most are all below the level of a Pentium 150. It doesn't take that much to get linux on them, just be smart about what you choose to install.

    Of course, that said, I also have to admit that it would be pretty nice if I didn't have to choose my packages carefully, because the defaults were ultra-slim. It does sometimes take a bit of work if the HD or RAM is too small (one of the funnier error messages I've ever gotten is "Memory size too small to load kernel--2M").

    So, props to these guys, and I'll give it a spin sometime, but for now, debian works wonders, and it's pretty well up to date.

  7. Don't see the need by moyix · · Score: 1 · on Linux on Older Hardware

    I've installed debian personally on a wide variety of low end machines:

    • Our main school server, colby.tjs.org, a P133, 64M of RAM, newer 20G HD. This runs DNS, POP3, SSH, Telnet, and Web without a hiccup. We came across a cheap IDE RAID card, and we're adding that in now.
    • A Pentium 75 which recently died from flaky hardware.
    • A Pentium 75 with all the spare SIMMS we could cram into it.
    • A 486, 14M of RAM, 275M HDD. It web serves.
    • A 386, specs forgotten since I accidentally killed it (oops!)

    In fact, the computers I work with most are all below the level of a Pentium 150. It doesn't take that much to get linux on them, just be smart about what you choose to install.

    Of course, that said, I also have to admit that it would be pretty nice if I didn't have to choose my packages carefully, because the defaults were ultra-slim. It does sometimes take a bit of work if the HD or RAM is too small (one of the funnier error messages I've ever gotten is "Memory size too small to load kernel--2M").

    So, props to these guys, and I'll give it a spin sometime, but for now, debian works wonders, and it's pretty well up to date.

  8. Speaking as someone who actually *owns* two.. by freshmkr · · Score: 1 · on The History Behind the Lisa UI

    Whatchew talkin' bout, Willis?

    I own two Lisas, a 2/10 and an original Lisa 1. I have never experienced an operating system crash, and the only time I have seen the office system tank (office system = finder and integrated office apps) was when a bug in one of my programs started eating up too much memory. When this occured, the Lisa put up a polite dialog box, saved everything in running apps, and rebooted itself. Beats a BSOD any day.
    <OSFLAME>

    Yes, the Lisa was a doomed project, but it wasn't poor design that made Apple take it off the market - not by a long shot. Here were the strikes against Lisa in 1984:

    • Cost. The Lisa started out costing $9,995 US, and prices never really dipped beneath $4,000 in later models. It should be noted that this price does not include the $3,000 ProFile external hard drive (5 MB!) for the Lisa 1 and Lisa 2/5.
    • Macintosh. Apple was making a cheaper, yet incompatible machine that was perceived as a 'baby Lisa' by many onlookers - and even Apple said that it had 'Lisa Technology' (i.e. a WIMP interface and 'Visual Fidelity' - IMHO a much better term for WYSIWYG). It was hard for buyers to justify the extra $3,000 or so.
    • Poor developer support. Apple sold Lisa language Workshops, where language includes Pascal, Clascal, C, BASIC, and COBOL. Unfortunately, the Workshops did not initially have support for Lisa Office System apps - their programs would have to be run from the Workshop or as their own shells (a shell is an operating environment run on startup - the Office System was a shell, as was the Workshop). Thus no office app integration or standard GUI for anything but Apple programs. Apple never did quite get around to finishing up the Lisa Toolkit, the programmers' library of standard Office System routines.
    • Speed. The Lisa was slow - it ran at only 5 MHz so that video accesses to memory could be interleaved with CPU accesses; it used a subset of the 68K instruction set to facilitate virtual memory and multitasking; and the OS and Office System (i.e. almost everything but the ROMs) was written in Pascal. Even writing a letter with a Lisa requires patience.
    But the Lisa had these attributes in its favor:
    • Memory protection/preemptive multitasking/virtual memory. These are exciting new technologies that will finally reach mainstream Mac users in sum with the release of MacOS X. See http://www.apple.com/macosx/inside.html
    • Robust file system. Inspired by the one at Xerox PARC, the Scavenger program automatically detects FS damage and fscks the disk. This may be part of the reason that my Lisa media has lasted so long.
    • Data sharing between Office System apps. By later versions of the Office System, graphs could be placed in text documents, terminal data could be pasted into spreadsheets, flowcharts could be copied into draw documents, etc. This is not news now, but in 1984...
    • Modular construction - all the low-voltage components of a Lisa (well, except for the speaker) can be accessed without a screwdriver.
    • Other niceties - soft power off, software contrast control, privacy dimmer (hit option-shift-keypad 0 to blank the screen), screen dimming after a preset time, session management (all open windows are noted at power off and restored at power on), and more.
    So the Lisa was not at all a poorly designed machine.

    Some sites for learning more about Lisa:

    --Tom Stepleton

  9. Speaking as someone who actually *owns* two.. by freshmkr · · Score: 1 · on The History Behind the Lisa UI

    Whatchew talkin' bout, Willis?

    I own two Lisas, a 2/10 and an original Lisa 1. I have never experienced an operating system crash, and the only time I have seen the office system tank (office system = finder and integrated office apps) was when a bug in one of my programs started eating up too much memory. When this occured, the Lisa put up a polite dialog box, saved everything in running apps, and rebooted itself. Beats a BSOD any day.
    <OSFLAME>

    Yes, the Lisa was a doomed project, but it wasn't poor design that made Apple take it off the market - not by a long shot. Here were the strikes against Lisa in 1984:

    • Cost. The Lisa started out costing $9,995 US, and prices never really dipped beneath $4,000 in later models. It should be noted that this price does not include the $3,000 ProFile external hard drive (5 MB!) for the Lisa 1 and Lisa 2/5.
    • Macintosh. Apple was making a cheaper, yet incompatible machine that was perceived as a 'baby Lisa' by many onlookers - and even Apple said that it had 'Lisa Technology' (i.e. a WIMP interface and 'Visual Fidelity' - IMHO a much better term for WYSIWYG). It was hard for buyers to justify the extra $3,000 or so.
    • Poor developer support. Apple sold Lisa language Workshops, where language includes Pascal, Clascal, C, BASIC, and COBOL. Unfortunately, the Workshops did not initially have support for Lisa Office System apps - their programs would have to be run from the Workshop or as their own shells (a shell is an operating environment run on startup - the Office System was a shell, as was the Workshop). Thus no office app integration or standard GUI for anything but Apple programs. Apple never did quite get around to finishing up the Lisa Toolkit, the programmers' library of standard Office System routines.
    • Speed. The Lisa was slow - it ran at only 5 MHz so that video accesses to memory could be interleaved with CPU accesses; it used a subset of the 68K instruction set to facilitate virtual memory and multitasking; and the OS and Office System (i.e. almost everything but the ROMs) was written in Pascal. Even writing a letter with a Lisa requires patience.
    But the Lisa had these attributes in its favor:
    • Memory protection/preemptive multitasking/virtual memory. These are exciting new technologies that will finally reach mainstream Mac users in sum with the release of MacOS X. See http://www.apple.com/macosx/inside.html
    • Robust file system. Inspired by the one at Xerox PARC, the Scavenger program automatically detects FS damage and fscks the disk. This may be part of the reason that my Lisa media has lasted so long.
    • Data sharing between Office System apps. By later versions of the Office System, graphs could be placed in text documents, terminal data could be pasted into spreadsheets, flowcharts could be copied into draw documents, etc. This is not news now, but in 1984...
    • Modular construction - all the low-voltage components of a Lisa (well, except for the speaker) can be accessed without a screwdriver.
    • Other niceties - soft power off, software contrast control, privacy dimmer (hit option-shift-keypad 0 to blank the screen), screen dimming after a preset time, session management (all open windows are noted at power off and restored at power on), and more.
    So the Lisa was not at all a poorly designed machine.

    Some sites for learning more about Lisa:

    --Tom Stepleton

  10. What an overexcited, uninformed post... by DoenerMord · · Score: 1 · on The History Behind the Lisa UI
    One of the biggest reasons I CANNOT stand the Macintrash is that (on the version of the Finder I used last) when you had a long enough file name, the end of it got hidden under the file size, and that you cannot (could not?) resize the individual columns.

    Gads man, calm down. This is the reason you don't like an operating system? When was the last time you used a Mac? I think this capability has been there since 8.0 or 8.5 atleast.

    On other topics, looks like the page has been slashdotted already. I like the other link to the Lisa page at http://galena.tjs.org/tom/. Someone mentioned they had a Lisa in a closet, anyone got one that still runs?

    -doenermord
    Don't blame the games, it takes a village to screw up a child.

  11. Troll! by profi · · Score: 1 · on Interview: Ask Steve Wozniak
    Apple had a preemptively multitasking OS with a pretty UI way back in 1983.

    Don't feel so special, Linux trolls. Implementing a buzzword-compliant Unix wannabe is almost trivial, and thousands of people do it every year in CS classes all over the world. The real problems start when you try to implement those features while staying backwards compatible with a legacy API. This is the real reason why it took Apple so long to come up with a replacement for MacOS, not the lack of focus or engineering talent.

  12. Questions re. development and early days of Apple by freshmkr · · Score: 1 · on Interview: Ask Steve Wozniak
    First question:
    If I understand the culture of the early days of Apple Computer correctly, there was very much the feeling among many of the employees that the small company was singlehandedly changing the world. It is clear now that what happened then shaped the personal computer industry for years to come. By comparison, today's computer industry seems dominated by large corporate concerns (e.g. Microsoft, IBM, Compaq, or even the present Apple), and the idea of a plucky little firm causing a revolution anywhere near the scale of Apple's and remaining commercially viable (c.f. Microsoft's war against Netscape) appears somewhat less likely than it used to. Are large corporations, the complexity of computers today, and/or the cost of innovation making it harder for small concerns to make big changes? This concerns me, as sometimes I doubt the ability of big business to do business with the best interests of progress in mind (not that Apple was anywhere near perfect, but...). Do you feel this way? Do you think Open Source may be an answer to this problem?

    Second question:
    I got into computers a bit too late to have the life-changing Apple ][ experience that so many people here have mentioned. Nevertheless, now that I style myself a "computer historian", I have one of my own and have learned how influential the machine was. During your tenure at Apple Computer, which computer were you most excited to have a hand in creating? If it was an Apple ][, which one? Was it because of how much fun it was to make, or the people you worked with, or how important you'd thought it would be?

    Third question:
    Did you have any part in the creation of Apple's Lisa computer? (it's my specialty, so I admit I've been curious for a while.)

    Thanks,
    --Tom