Re: [puredyne] Motion detection packages
Hey James, Yay, very nice, looks good. Care to share the patch? Julian On 15 September 2011 01:07, James Harkins jamshar...@gmail.com wrote: At Mon, 12 Sep 2011 15:02:46 +0200, Benjamin ~ 01xy wrote: maybe another way to do it is to use [pix_crop] object in GEM as in this example (motion checked by a color approach, it can also be done with [pix_movement]) http://yamatierea.org/papatchs/#detc-mvt or pidip object [pdp_mgrid] (see example in pdextended) of course, opencv objects are also really efficient for some kinds of motion detection Cool, thanks. Anyway, the last approach I was thinking of (motion detection slicing instead of the other way around) turned out to work marvelously. http://www.dewdrop-world.net/words/motion_detect_small.avi Screenshot of the pd patch is on my blog (link is in the signature). Thanks for the suggestions, all -- James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
At Mon, 12 Sep 2011 15:02:46 +0200, Benjamin ~ 01xy wrote: maybe another way to do it is to use [pix_crop] object in GEM as in this example (motion checked by a color approach, it can also be done with [pix_movement]) http://yamatierea.org/papatchs/#detc-mvt or pidip object [pdp_mgrid] (see example in pdextended) of course, opencv objects are also really efficient for some kinds of motion detection Cool, thanks. Anyway, the last approach I was thinking of (motion detection slicing instead of the other way around) turned out to work marvelously. http://www.dewdrop-world.net/words/motion_detect_small.avi Screenshot of the pd patch is on my blog (link is in the signature). Thanks for the suggestions, all -- James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
Hi James, Bit late on this but I have had good results from William Brent's 'pix_motion_sector' http://williambrent.conflations.com/pages/research.html Cheers, Julian On 12 September 2011 02:29, James Harkins jamshar...@gmail.com wrote: At Sun, 11 Sep 2011 13:05:31 -0300, palmieri, ricardo wrote: and this? http://www.hangar.org/wikis/lab/doku.php?id=start:puredata_opencv Oh, interesting. Thanks (and thanks to Marko for the other suggestions). I got another idea this morning -- instead of slicing the frame and running a bunch of motion detectors, run one motion_detection object on the whole frame and slice that result. The problem was that motion detection is stateful (depends on previous frames) so I would have to have a separate object for each slice, but the [#moment] calculation is not. That should simplify the patch a lot. James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
Hello, maybe another way to do it is to use [pix_crop] object in GEM as in this example (motion checked by a color approach, it can also be done with [pix_movement]) http://yamatierea.org/papatchs/#detc-mvt or pidip object [pdp_mgrid] (see example in pdextended) of course, opencv objects are also really efficient for some kinds of motion detection best++ Benjamin James Harkins a écrit : At Sun, 11 Sep 2011 13:05:31 -0300, palmieri, ricardo wrote: and this? http://www.hangar.org/wikis/lab/doku.php?id=start:puredata_opencv Oh, interesting. Thanks (and thanks to Marko for the other suggestions). I got another idea this morning -- instead of slicing the frame and running a bunch of motion detectors, run one motion_detection object on the whole frame and slice that result. The problem was that motion detection is stateful (depends on previous frames) so I would have to have a separate object for each slice, but the [#moment] calculation is not. That should simplify the patch a lot. James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
and this? http://www.hangar.org/wikis/lab/doku.php?id=start:puredata_opencv 2011/9/11 Marko Cebokli s57...@hamradio.si Maybe these could be of use: http://www.lavrsen.dk/foswiki/bin/view/Motion/WebHome http://www.zoneminder.com/ Marko Cebokli On Sunday 11 September 2011 05:30:48 Jonathan Wilkes wrote: - Original Message - From: James Harkins jamshar...@gmail.com To: puredyne puredyne@goto10.org Cc: Sent: Saturday, September 10, 2011 10:58 PM Subject: [puredyne] Motion detection packages I'm starting to look into some motion detection techniques. I got some distance with pure data + gridflow (the [#motion_detection] object behaves quite nicely -- it takes the difference between frames but includes some other tweaks for a cleaner signal). But what I really wanted to do was divide the image into nine blocks, run motion detection independently on each one, and have information about where in the frame the motion is happening and also movement from region to region. That became a problem, largely because gridflow's documentation has to rank among the worst I've ever seen. (Such as: Inlet 2: crosshairs. Oh, very nice. Yes, out of the thousands of possible things that could mean, we're to psychically divine the one that is actually correct.) Unfortunately, that's fairly verbose documentation for a pure data external. (But it looks to me like it's Inlet 4-- which version of Gridflow are you using?) You might try posting this on the pd-list at pd-l...@iem.at . The GF author reads/posts there and can probably flesh out the documentation directly. So I'm looking for recommendations on other open-source approaches. OpenCV+processing maybe, but for some reason willowgarage.com is blocked on the mainland (so I have to rely on a proxy that isn't always stable). That means learning another computer language on short order, not sure I have time. Can anybody think of something that would be closer to what I have in mind out of the box? Just to save me some development time reinventing the wheel. Thanks, James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne -- twitt-me: @ricardopalmieri mobile# +551185833173 [memelab.com.br] [palm.estudiolivre.org] [myspace.com/livenoisetupi] [skype:palmieriricardo] [msn: ricardopalmi...@bol.com.br] [linux user # 392484] --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
At Sun, 11 Sep 2011 13:05:31 -0300, palmieri, ricardo wrote: and this? http://www.hangar.org/wikis/lab/doku.php?id=start:puredata_opencv Oh, interesting. Thanks (and thanks to Marko for the other suggestions). I got another idea this morning -- instead of slicing the frame and running a bunch of motion detectors, run one motion_detection object on the whole frame and slice that result. The problem was that motion detection is stateful (depends on previous frames) so I would have to have a separate object for each slice, but the [#moment] calculation is not. That should simplify the patch a lot. James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
- Original Message - From: James Harkins jamshar...@gmail.com To: puredyne puredyne@goto10.org Cc: Sent: Saturday, September 10, 2011 10:58 PM Subject: [puredyne] Motion detection packages I'm starting to look into some motion detection techniques. I got some distance with pure data + gridflow (the [#motion_detection] object behaves quite nicely -- it takes the difference between frames but includes some other tweaks for a cleaner signal). But what I really wanted to do was divide the image into nine blocks, run motion detection independently on each one, and have information about where in the frame the motion is happening and also movement from region to region. That became a problem, largely because gridflow's documentation has to rank among the worst I've ever seen. (Such as: Inlet 2: crosshairs. Oh, very nice. Yes, out of the thousands of possible things that could mean, we're to psychically divine the one that is actually correct.) Unfortunately, that's fairly verbose documentation for a pure data external. (But it looks to me like it's Inlet 4-- which version of Gridflow are you using?) You might try posting this on the pd-list at pd-l...@iem.at . The GF author reads/posts there and can probably flesh out the documentation directly. So I'm looking for recommendations on other open-source approaches. OpenCV+processing maybe, but for some reason willowgarage.com is blocked on the mainland (so I have to rely on a proxy that isn't always stable). That means learning another computer language on short order, not sure I have time. Can anybody think of something that would be closer to what I have in mind out of the box? Just to save me some development time reinventing the wheel. Thanks, James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne
Re: [puredyne] Motion detection packages
Maybe these could be of use: http://www.lavrsen.dk/foswiki/bin/view/Motion/WebHome http://www.zoneminder.com/ Marko Cebokli On Sunday 11 September 2011 05:30:48 Jonathan Wilkes wrote: - Original Message - From: James Harkins jamshar...@gmail.com To: puredyne puredyne@goto10.org Cc: Sent: Saturday, September 10, 2011 10:58 PM Subject: [puredyne] Motion detection packages I'm starting to look into some motion detection techniques. I got some distance with pure data + gridflow (the [#motion_detection] object behaves quite nicely -- it takes the difference between frames but includes some other tweaks for a cleaner signal). But what I really wanted to do was divide the image into nine blocks, run motion detection independently on each one, and have information about where in the frame the motion is happening and also movement from region to region. That became a problem, largely because gridflow's documentation has to rank among the worst I've ever seen. (Such as: Inlet 2: crosshairs. Oh, very nice. Yes, out of the thousands of possible things that could mean, we're to psychically divine the one that is actually correct.) Unfortunately, that's fairly verbose documentation for a pure data external. (But it looks to me like it's Inlet 4-- which version of Gridflow are you using?) You might try posting this on the pd-list at pd-l...@iem.at . The GF author reads/posts there and can probably flesh out the documentation directly. So I'm looking for recommendations on other open-source approaches. OpenCV+processing maybe, but for some reason willowgarage.com is blocked on the mainland (so I have to rely on a proxy that isn't always stable). That means learning another computer language on short order, not sure I have time. Can anybody think of something that would be closer to what I have in mind out of the box? Just to save me some development time reinventing the wheel. Thanks, James -- James Harkins /// dewdrop world jamshar...@dewdrop-world.net http://www.dewdrop-world.net Come said the Muse, Sing me a song no poet has yet chanted, Sing me the universal. -- Whitman blog: http://www.dewdrop-world.net/words audio clips: http://www.dewdrop-world.net/audio more audio: http://soundcloud.com/dewdrop_world/tracks --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne --- Puredyne@goto10.org http://identi.ca/group/puredyne irc://irc.goto10.org/puredyne