Menu
▾
▴
motion-user
|
From: Will S. <wil...@gm...> - 2013-11-25 16:30:51
|
semaphore45 asked, > I guess in that context local storage stands for local copy? I my case, that would mean if the network were down, any video associated with a detected motion would be stored locally on the camera 'unit', which in the case of the PI camera, would be on the Raspberry's 8G flashdrive (until the network came back up). Will On Mon, Nov 25, 2013 at 6:43 AM, <sem...@ya...> wrote: > I guess in that context local storage stands for local copy? > > Le 25 nov. 2013 à 06:32, Will Stewart a écrit : > > Steve wrote; > > > offload pedestrian detection to a more powerful platform > > I'm taking a like approach. I have different cameras, each with their own > capabilities, resolutions, on-board processing abilities, etc (e.g., PI > camera vs. Axis, for an example), so my solution will be heterogeneous. And > some in OpenHab want to be able to use h.264, so I'm exploring that as well > (which takes me out of the Motion domain for that format). > > Can you share your approach to pedestrian detection, i.e, is it using > feature detection, cascade classifier, basic shape attributes, etc? > > > The idea that you should transmit video streams over the network 24/7 > is a waste of bandwidth > > Agreed, I'm only interested in transmitting motion events, and on the > cameras with sufficient onboard processing, only motion events that pass a > particular criteria test. I'm thinking of storing 0.5 fps when we are away, > though have not finalized that yet. > > > with local storage you are protected from network outages. > > Exactly. > > Best, > > Will > > > > On Sun, Nov 24, 2013 at 8:25 PM, Steve Goldsmith <sg...@gm...> wrote: > >> I have the VM host (think personal micro cloud) for other purposes >> (Tomcat instances, wiki, blog Jenkins, Nexus, etc.), but having the spare >> virtual CPUs allows me to offload pedestrian detection to a more powerful >> platform. Obviously you could use a quad core Android stick loaded with >> Linux and handle a lot of what I'm doing with the server. My idea is to use >> the quad core to handle the people detection real time since I use the dual >> core MK808 as a camera now to do motion detection, Xvid compression and >> SCP. You can also stash one of these out of site as a file server to secure >> your video file if your computers or cameras are stolen. >> >> There's not right way to do it. You can go balls to the wall and buy $500 >> HD h.264 cameras, powerful servers, etc. or build cameras like I've done >> for about the same as a decent 640x480 MJPEG camera. I like the distributed >> idea though because the camera does most of the heavy lifting and with >> local storage you are protected from network outages. The idea that you >> should transmit video streams over the network 24/7 is a waste of >> bandwidth. How much are you actually recording? >> >> I'm also using a dedicated $25 wireless N router, so all the cameras are >> on a separate network from my main WAN/LAN. This prevents 12 Mbps of >> wireless traffic from going over my main LAN. >> >> >> On Sun, Nov 24, 2013 at 12:26 PM, <sem...@ya...> wrote: >> >>> Seems like a pretty heavy setup, although without an illustration I can >>> only imagine what goes where. At least two machines, how many smart cams, >>> two independant routers? Built from what? How is Xvid configured? I read >>> long ago that it was quite difficult to get a decent quality while keeping >>> CPU usage and storage reasonable. >>> >>> >> >> >> ------------------------------------------------------------------------------ >> Shape the Mobile Experience: Free Subscription >> Software experts and developers: Be at the forefront of tech innovation. >> Intel(R) Software Adrenaline delivers strategic insight and game-changing >> conversations that shape the rapidly evolving mobile landscape. Sign up >> now. >> >> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk >> _______________________________________________ >> Motion-user mailing list >> Mot...@li... >> https://lists.sourceforge.net/lists/listinfo/motion-user >> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >> >> > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up > now. > > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk_______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > > > > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up > now. > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk > _______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > > |
|
From: C.T. <sem...@ya...> - 2013-11-25 17:34:01
|
And what kind of performance do you get from this smart camera? Does it has night vision? Can e raspberry support more than one cam? Enviado desde mi iPad El 25/11/2013, a las 11:30, Will Stewart <wil...@gm...> escribió: > semaphore45 asked, > > > I guess in that context local storage stands for local copy? > > I my case, that would mean if the network were down, any video associated with a detected motion would be stored locally on the camera 'unit', which in the case of the PI camera, would be on the Raspberry's 8G flashdrive (until the network came back up). > > Will > > > On Mon, Nov 25, 2013 at 6:43 AM, <sem...@ya...> wrote: >> I guess in that context local storage stands for local copy? >> >> Le 25 nov. 2013 à 06:32, Will Stewart a écrit : >> >>> Steve wrote; >>> >>> > offload pedestrian detection to a more powerful platform >>> >>> I'm taking a like approach. I have different cameras, each with their own capabilities, resolutions, on-board processing abilities, etc (e.g., PI camera vs. Axis, for an example), so my solution will be heterogeneous. And some in OpenHab want to be able to use h.264, so I'm exploring that as well (which takes me out of the Motion domain for that format). >>> >>> Can you share your approach to pedestrian detection, i.e, is it using feature detection, cascade classifier, basic shape attributes, etc? >>> >>> > The idea that you should transmit video streams over the network 24/7 is a waste of bandwidth >>> >>> Agreed, I'm only interested in transmitting motion events, and on the cameras with sufficient onboard processing, only motion events that pass a particular criteria test. I'm thinking of storing 0.5 fps when we are away, though have not finalized that yet. >>> >>> > with local storage you are protected from network outages. >>> >>> Exactly. >>> >>> Best, >>> >>> Will >>> >>> >>> >>> On Sun, Nov 24, 2013 at 8:25 PM, Steve Goldsmith <sg...@gm...> wrote: >>>> I have the VM host (think personal micro cloud) for other purposes (Tomcat instances, wiki, blog Jenkins, Nexus, etc.), but having the spare virtual CPUs allows me to offload pedestrian detection to a more powerful platform. Obviously you could use a quad core Android stick loaded with Linux and handle a lot of what I'm doing with the server. My idea is to use the quad core to handle the people detection real time since I use the dual core MK808 as a camera now to do motion detection, Xvid compression and SCP. You can also stash one of these out of site as a file server to secure your video file if your computers or cameras are stolen. >>>> >>>> There's not right way to do it. You can go balls to the wall and buy $500 HD h.264 cameras, powerful servers, etc. or build cameras like I've done for about the same as a decent 640x480 MJPEG camera. I like the distributed idea though because the camera does most of the heavy lifting and with local storage you are protected from network outages. The idea that you should transmit video streams over the network 24/7 is a waste of bandwidth. How much are you actually recording? >>>> >>>> I'm also using a dedicated $25 wireless N router, so all the cameras are on a separate network from my main WAN/LAN. This prevents 12 Mbps of wireless traffic from going over my main LAN. >>>> >>>> >>>> On Sun, Nov 24, 2013 at 12:26 PM, <sem...@ya...> wrote: >>>>> Seems like a pretty heavy setup, although without an illustration I can only imagine what goes where. At least two machines, how many smart cams, two independant routers? Built from what? How is Xvid configured? I read long ago that it was quite difficult to get a decent quality while keeping CPU usage and storage reasonable. >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> Shape the Mobile Experience: Free Subscription >>>> Software experts and developers: Be at the forefront of tech innovation. >>>> Intel(R) Software Adrenaline delivers strategic insight and game-changing >>>> conversations that shape the rapidly evolving mobile landscape. Sign up now. >>>> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk >>>> _______________________________________________ >>>> Motion-user mailing list >>>> Mot...@li... >>>> https://lists.sourceforge.net/lists/listinfo/motion-user >>>> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >>> >>> ------------------------------------------------------------------------------ >>> Shape the Mobile Experience: Free Subscription >>> Software experts and developers: Be at the forefront of tech innovation. >>> Intel(R) Software Adrenaline delivers strategic insight and game-changing >>> conversations that shape the rapidly evolving mobile landscape. Sign up now. >>> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk_______________________________________________ >>> Motion-user mailing list >>> Mot...@li... >>> https://lists.sourceforge.net/lists/listinfo/motion-user >>> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >> >> >> ------------------------------------------------------------------------------ >> Shape the Mobile Experience: Free Subscription >> Software experts and developers: Be at the forefront of tech innovation. >> Intel(R) Software Adrenaline delivers strategic insight and game-changing >> conversations that shape the rapidly evolving mobile landscape. Sign up now. >> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk >> _______________________________________________ >> Motion-user mailing list >> Mot...@li... >> https://lists.sourceforge.net/lists/listinfo/motion-user >> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up now. > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk > _______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome |
|
From: Steve G. <sg...@gm...> - 2013-11-25 17:52:26
|
I can drive a 720P HD USB webcam @ 20 FPS using about 40% of the CPU. You typically want to keep the total CPU < 50% for motion detection and recording. The rest of the bandwidth is used for secondary compression, SCP, etc. As far as the Pi goes I'm not the expert, but with a single core you are going to be limited unless you rely on hardware encode/decode of the video stream. It's a USB camera, so if it has night vision, then yes of course. On Mon, Nov 25, 2013 at 12:33 PM, C.T. <sem...@ya...> wrote: > And what kind of performance do you get from this smart camera? Does it > has night vision? Can e raspberry support more than one cam? > > |
|
From: Will S. <wil...@gm...> - 2013-11-25 20:20:09
|
C.T. asked, > And what kind of performance do you get from this smart camera? Does it has night vision? Can e raspberry support more than one cam? The Raspberry PI with Camera is basically an ARM6 with a 5MP camera plugin. There is an IR version <http://www.raspberrypi.org/archives/5089>available. I'm only looking to use 1 camera per Raspberry, where it will be a single camera unit that runs Motion (and perhaps other software), sending video only during a motion event, or when I otherwise want to see or record video. See an example <http://www.raspberrypi.org/archives/5071>. It doesn't cover everyone's feature set wishlist or every possible situation by any stretch of the imagination, but with a $35 processor board+Linux, a $25 5MP camera plug-in, and a $7 dummy camera enclosure, this can be one inexpensive surveillance camera that could meet many people's needs. YMMV Surveillance design has been trending for years away from a purely centralized host to more distributed processing, where cameras with processors could focus on determining motion (and other) events themselves, then sending the high rate video only during those times to moderate the amount of network traffic. Many out-of-the-box camera motion detectors that come with inexpensive cameras are rudimentary and may only have fairly coarse adjustments at best. Motion provides many exceptional features, starting with masking, that help to reduce false alarms, though there are other factors I want to consider as well. For example, at 3am a standard video motion detection algorithm might trigger on my dog at 5m (which I don't care about) the same as it would a person at 10m. So my solution could roughly be described by these steps; 1. Detect motion (Motion for MJPEG, OpenCV for H264) 2. Extract object features (OpenCV) 3. Classify object (OpenCV) 4. Look at the current mode's rules to determine whether or not to alarm (OpenHab) . There will likely be times when we want to detect the faces of friends if they stop by when we are not at home, and either give them a personalized automated greeting at the front door speaker, or SMS us with their name so that we could converse with them via 2-way audio from our phone. This would also work for unknown persons showing up at the front door in any situation. I am looking at the PI camera to perform the 1st step at a minimum (that may be all it can do, barring more umph from the gpu that I am investigating). Steve's ROI project looks very intriguing, so I'm going to dig into that later this week. Cheers, Will On Mon, Nov 25, 2013 at 12:33 PM, C.T. <sem...@ya...> wrote: > And what kind of performance do you get from this smart camera? Does it > has night vision? Can e raspberry support more than one cam? > > Enviado desde mi iPad > > El 25/11/2013, a las 11:30, Will Stewart <wil...@gm...> > escribió: > > semaphore45 asked, > > > I guess in that context local storage stands for local copy? > > I my case, that would mean if the network were down, any video associated > with a detected motion would be stored locally on the camera 'unit', which > in the case of the PI camera, would be on the Raspberry's 8G flashdrive > (until the network came back up). > > Will > > > On Mon, Nov 25, 2013 at 6:43 AM, <sem...@ya...> wrote: > >> I guess in that context local storage stands for local copy? >> >> Le 25 nov. 2013 à 06:32, Will Stewart a écrit : >> >> Steve wrote; >> >> > offload pedestrian detection to a more powerful platform >> >> I'm taking a like approach. I have different cameras, each with their own >> capabilities, resolutions, on-board processing abilities, etc (e.g., PI >> camera vs. Axis, for an example), so my solution will be heterogeneous. And >> some in OpenHab want to be able to use h.264, so I'm exploring that as well >> (which takes me out of the Motion domain for that format). >> >> Can you share your approach to pedestrian detection, i.e, is it using >> feature detection, cascade classifier, basic shape attributes, etc? >> >> > The idea that you should transmit video streams over the network 24/7 >> is a waste of bandwidth >> >> Agreed, I'm only interested in transmitting motion events, and on the >> cameras with sufficient onboard processing, only motion events that pass a >> particular criteria test. I'm thinking of storing 0.5 fps when we are away, >> though have not finalized that yet. >> >> > with local storage you are protected from network outages. >> >> Exactly. >> >> Best, >> >> Will >> >> >> >> On Sun, Nov 24, 2013 at 8:25 PM, Steve Goldsmith <sg...@gm...>wrote: >> >>> I have the VM host (think personal micro cloud) for other purposes >>> (Tomcat instances, wiki, blog Jenkins, Nexus, etc.), but having the spare >>> virtual CPUs allows me to offload pedestrian detection to a more powerful >>> platform. Obviously you could use a quad core Android stick loaded with >>> Linux and handle a lot of what I'm doing with the server. My idea is to use >>> the quad core to handle the people detection real time since I use the dual >>> core MK808 as a camera now to do motion detection, Xvid compression and >>> SCP. You can also stash one of these out of site as a file server to secure >>> your video file if your computers or cameras are stolen. >>> >>> There's not right way to do it. You can go balls to the wall and buy >>> $500 HD h.264 cameras, powerful servers, etc. or build cameras like I've >>> done for about the same as a decent 640x480 MJPEG camera. I like the >>> distributed idea though because the camera does most of the heavy lifting >>> and with local storage you are protected from network outages. The idea >>> that you should transmit video streams over the network 24/7 is a waste of >>> bandwidth. How much are you actually recording? >>> >>> I'm also using a dedicated $25 wireless N router, so all the cameras are >>> on a separate network from my main WAN/LAN. This prevents 12 Mbps of >>> wireless traffic from going over my main LAN. >>> >>> >>> On Sun, Nov 24, 2013 at 12:26 PM, <sem...@ya...> wrote: >>> >>>> Seems like a pretty heavy setup, although without an illustration I can >>>> only imagine what goes where. At least two machines, how many smart cams, >>>> two independant routers? Built from what? How is Xvid configured? I read >>>> long ago that it was quite difficult to get a decent quality while keeping >>>> CPU usage and storage reasonable. >>>> >>>> >>> >>> >>> ------------------------------------------------------------------------------ >>> Shape the Mobile Experience: Free Subscription >>> Software experts and developers: Be at the forefront of tech innovation. >>> Intel(R) Software Adrenaline delivers strategic insight and game-changing >>> conversations that shape the rapidly evolving mobile landscape. Sign up >>> now. >>> >>> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk >>> _______________________________________________ >>> Motion-user mailing list >>> Mot...@li... >>> https://lists.sourceforge.net/lists/listinfo/motion-user >>> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >>> >>> >> >> ------------------------------------------------------------------------------ >> Shape the Mobile Experience: Free Subscription >> Software experts and developers: Be at the forefront of tech innovation. >> Intel(R) Software Adrenaline delivers strategic insight and game-changing >> conversations that shape the rapidly evolving mobile landscape. Sign up >> now. >> >> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk_______________________________________________ >> Motion-user mailing list >> Mot...@li... >> https://lists.sourceforge.net/lists/listinfo/motion-user >> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >> >> >> >> >> ------------------------------------------------------------------------------ >> Shape the Mobile Experience: Free Subscription >> Software experts and developers: Be at the forefront of tech innovation. >> Intel(R) Software Adrenaline delivers strategic insight and game-changing >> conversations that shape the rapidly evolving mobile landscape. Sign up >> now. >> >> http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk >> _______________________________________________ >> Motion-user mailing list >> Mot...@li... >> https://lists.sourceforge.net/lists/listinfo/motion-user >> http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome >> >> > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up > now. > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk > > _______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > > > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up > now. > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk > _______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > > |
|
From: tosiara <to...@gm...> - 2013-12-05 14:20:16
|
Github shows there pretty many people still improving motion: https://github.com/sackmotion/motion/network/members I beleive some day one good guy will merge all the good stuff and release motion v.4. Or even it will be Kenneth itself ;) On Thu, Nov 21, 2013 at 6:40 PM, Grant <ema...@gm...> wrote: > There hasn't been a motion release in quite a while. Is the project > dead? Is there a good alternative program? > > - Grant > > > ------------------------------------------------------------------------------ > Shape the Mobile Experience: Free Subscription > Software experts and developers: Be at the forefront of tech innovation. > Intel(R) Software Adrenaline delivers strategic insight and game-changing > conversations that shape the rapidly evolving mobile landscape. Sign up > now. > http://pubads.g.doubleclick.net/gampad/clk?id=63431311&iu=/4140/ostg.clktrk > _______________________________________________ > Motion-user mailing list > Mot...@li... > https://lists.sourceforge.net/lists/listinfo/motion-user > http://www.lavrsen.dk/twiki/bin/view/Motion/WebHome > |
Thanks for helping keep SourceForge clean.
X