buellwinkle

To log the plates, you need low resolution. We use 1MP cameras but run them at 640x480. The reason is the larger the image, the harder the CPU has to work to translate the image to text. Also, we run 5 fps and that seems to be good to 30-40mph, you may need slightly higher rate to get 65mph. You want the least you can get away. Ballpark, we use about a 7-8' wide field of view as the plate passes the area of detection. Most LPR software requires the plate numbers be a certain percentage of the image, but in general terms, with VGA resolution, the plate numbers should be about 15' pixels tall. This requires a telephoto lens. For example, we capture plates with one camera at about 70' and that requires a 50mm lens. We have one in the works where we plan on using a 22mm lens to capture plates at 25-30'. What's your budget for software? We use Milestone XProtect LPR, runs $300 per server + $1,295 per camera + the NVR license cost (minimum is Express edition which is $99/cam). This is middle of the road pricing. High end LPR can be $2,500-5,000 per camera. Low end would be like Geovision LPR, runs about $800/cam + NVR software fee. I have not tried Geovision, but didn't get high praise. There's an open source project at www.openalpr.com that's very cool, but you have to write your own C++ code, the provide the engine to do the plate translation using open source projects. So that would be free. ACTi has a free LPR program, have not tried it, but it's free, but likely only works with their cams. I didn't go that route because plate list match is not available, but I believe it logs the plate numbers. There is a company working on embedding the LPR translation in certain cameras by running their software on the camera's Linux. IPConfigure does it but it doesn't work all that well. This is from Carmen that sells the engine now to different companies and is the basis for Milestone LPR. They showed it off at ISC West last year and was very cool, but was not a product yet. Carmen is like OpenALPR in that they provide an engine, you write the code, but Carmen is a commercial solution, not free.

What is your goal? Do you want to log license plate numbers, do you want to create an event based on a license plate match, do you want to just be able to read a plate number in video?

If that's his trade, then Anixter would be the wholesale place to get it from as they seem to so a lot of that sort of stuff plus surveillance cameras. They have offices in all major cities in the U.S. You can even go to their website and grab the forms to fill out to get started. They usually require a contractor's license or reseller permit to open the account, but sure he has that.

Just cut the male RJ45 plug off, slip the pieces you got on that end. Test fit to make sure you have put it on in the correct direction. Then when you are happy, crimp on a new RJ45 plug. You can get tools to crimp, test as well as a packet of RJ45 plugs for pretty cheap online, maybe $20 for everything. Plenty of videos on youtube on how to do it. I don't use them because it doesn't rain in So Cal.

Just set the max shutter speed to about 1/200 of a second and start from there. Depending on your camera, you may have to tweak those values. Also aim the camera about 30 degrees to path of travel. Too steep an angle and you won't be able to read the plate, to small an angle and the lead car will block a following car and miss that plate.

We run BlueIris and Milestone XProtect on the same very old PC (2009) that was cheap and slow when new at one site and no problems. Just make sure when you mix solutions you are weary of port numbers as both can't be port 80.

Have you tried upgrading or downgrading the firmware using the config tool? Can you see the camera in config tool? Have you tried telneting into the camera. Also, does TFTP work for their cameras (I know it works for their DVRs).

First, nobody I know of makes an LPR camera other than Axis with iPConfigure software that actually does LPR in camera. LPR is License Plate Recognition where OCR software is used to translate a plate number to text. Most so called "LPR" cameras are nothing more than an ordinary camera with a very wide range varifocal lens and low light sensitivity. I've seen images from the Geovision LPR camera and I have been doing LPR for a while now, I can tell you that camera is not capable of true LPR. It's more of an LPC (License Plate Capture) camera, meaning you will be able to see a plate in the video, but not have enough clarity for an OCR software to actually read that plate number, none that I know of. So how do these LPC cameras differ from ordinary cameras. The Geovision has a max shutter speed of 1/500 of a second and has a few IR LEDs. It depends on the reflectivity of the plate to shine and be seen despite headlights or taillights. The images I've seen were very dark and the plate barely visible but readable. It has a wide range varifocal lens so you can set it up with a close-up of the plate number for better readability. I've used Hikvision 3MP cameras with a telephoto lens and gotten better results. We use an Axis q1604 with a 5-50mm lens and we read plates using LPR software fairly reliably and way better clarity than the Geovision LPR camera. If your goal is just LPC, then most cameras setup properly with the right lens can do that. By me, the toll roads folks implemented LPR recently to collect tolls. They do not use IR, they use flash photography which makes the most sense. There are some state with red numbers on license plates, like Illinois, Massachusetts, Oklahoma that will not work with IR.

You can reset the camera by pressing the reset button as you have and hold it while connecting the camera to power (PoE or 12V) and hold it for about 10 seconds. After the camera boots up, you should see it in SADP with an IP of 192.0.0.64 and you can change it there. If you don't see it in SADP, it could be Winpcap did not get installed properly, try installing the latest version separately from the Winpcap website.

I pm'ed you in case you want to send me the camera, I can reset it with whatever firmware you want in English.

I use an Intel NUC i5, 4GB RAM, 2TB drive, 10 cameras (8 are 3MP, 2 are 1.3MP) and it works great for me. For 4, I would save the money and get the Intel NUC Celeron, 4GB RAM, 1TB drive, should cost you under $250 for the hardware. I like the NUC's because they are tiny (4.5" square, 2" tall), they use very low power CPU, so minimal heat, heck, no CPU fan to burn out and mine uses about 12-14W. I also use efficient software, in my case Milestone XProtect. You can get the free Go version as it supports 8 cameras with 5 days recording or pay $49/cam without those limitations for the Essential version. Or many times, you can get pretty decent software from the camera company for free, like ACTi NVR 3.0, Hikvision iVMS4200 PCNVR or Dahua iDMSS. The downside to the NUC is only one drive, so you can't have redundancy in a RAID configuration. If that's important to you, look at the Dell T20, can hold 4 3.5" drives + 2 2.5" and has beefy fans, well made, cheap. The Celeron is fine for 4 cameras, more than that, the Xeon version works well, we run it with about 20 cameras. As for packaged system, do you mean a DVR/NVR or a pre-built system from major NVR software company like a Milestone appliance? If installing Windows and Windows based software is beyond your skills, then this is the way to go. If you don't have issues with Windows, then once you install the software, setting up cameras, motion detection, remote access and the like is the same level of effort.

I differentiate software engineer as someone that work on lower level code, in my case operating systems vs. application code, but done that too. Never in 35 years did I ever have the title of programmer, I would have done something else rather than endure that, LOL. But outgrew that, no longer satisfied with "engineer" in my title, I've grown to scientist, a notch above an engineer, only because there's a civil engineer that I know that's so incompetent, I refuse to be called an engineer anymore. Marketing hype or not, the purple is the same price as their desktop drives and right now the replacement drive is sitting on my desktop, hence it's now a desktop drive aka. paperweight.

Thanks for clarifying that, a source of confusion since the camera came out. Like I said, using a standard chart, I saw no difference in optical resolution between this 3MP and other ones I had laying around, if anything, the Hik was slightly better.

Not sure how to do it in XProtect, but most cameras accept an URL command to reboot, so you can write a Windows script that issues that command to each cameras, then schedule it with Windows Task Scheduler. I have script I wrote that checks each camera by making an HTTP request and emails me if it's not responding. If you have a managed PoE switch, you may be able to send it a command to reboot itself, making it power cycle all the cameras. May be better as it will power cycle a hung camera where sending the camera a command to reboot to a camera that's not responsive may be moot. The other thing you can try if your PoE switch is not managed is to get a Wemo, plug the switch in that and then you can reboot it remotely but not sure if you can schedule it.

There actually is, but only seen it on crime shows like CSI. Maybe if you call the network they can share the technology they use. I'm still waiting for CSI to get back to me on how they do this -

You can use surface mounted conduit and jbox instead of drilling holes, but then it becomes an appearance issue. If you sell the house, either leave the cameras or putty over the hole. Why do you need a 2" hole, at most a 1" hole is needed.

On the camera's web interface, when that feature is turned on, you'll see bright green grids appear at the spot where motion is detected. Does not show up in the stream, so you likely won't see it in the NVR itself.

Actually, maybe I started off too low end for you. Our actual LPR camera we use is an Axis Q1604 with a Fujinon 5-50mm lens inside an Axis outdoor enclosure (don't get the outdoor "-E" model as the lens won't fit in that enclosure). We combined it with a Raytec RM100-30 illuminator to allow for a full view of the car at night that overcomes the license plate reflection and taillights. We are looking at putting in an Avigilon dome with a 9-22mm lens for our next project where a vandal dome is needed because of the low install height and there's not too many vandal domes with telephoto lenses. Have you seen the article in my blog on license plate recognition?

Here's a picture of the camera in my hand. There is a smaller dome that's very cool looking from Avigilon - http://avigilon.com/products/video-surveillance/cameras/hd-micro-dome/hd-micro-dome-cameras/2-mp-hd-micro-dome/ As for NVR's, they are sold by number of cameras, has nothing to do with it's capabilities as many times a 4, 8 and 16 channel NVR are the exact same device but only lets you configure the cameras you are licensed for regardless of where they sit on your network. I personally prefer a PC running Windows with NVR software. I record everything based on motion detection.

The title for the review says "outdoor", just saying. But yes, it's outdoors in the front of my home now, was going to take it down today and ready for the next camera. I'll take a picture of it next to something so you can relate to it's size.

What I've put in for small business is a small Dahua NVR with PoE and 4 Dahua 1080P mini domes or bullets, usually runs about $129USD per camera, $269 for the NVR, $60 for a 1TB HDD. Then the question is VAT and other fees added by your country. This would be the best bang for the buck and does not include wiring and installation, but is plug and play for the most part. Many NVR's are not capable of viewing more than 1080P as is with most monitors , so 3MP will cost more and may not give you better returns. Also, many NVR's assume 16:9 aspect ratio, so using 3MP which is 4:3 will give you a smaller display as it will be cut to fit. Don't get me wrong, I love 3MP and all mine are set that way, but I'm not using an NVR.

We use Milestone XProtect Express with about 21 cameras, mostly 3MP not even gigabit switches, but several 100Mb switches feeding up to a gigabit uplink using all sub streams and displaying that many cameras using their SmartClient or their WebClient works fine. If I try and use all mainstreams on a modern day 4th gen i7, I would hit 100% CPU making it bad. I can still playback the recorded main stream, but even then, I would have "views" of small groups of cameras for a location so as not to overwhelm the PC I'm viewing it on when playing back the recorded video.

Did not know you gave up on CK years ago. Maybe just paranoid, but I'm worried about the NAND flash even in the camera because I have an Arduino with a temp sensor updating the OSD for a camera with temperature readings on a web cam. Just a trait of NAND flash that includes SD cards, SSD and flash chips in cameras. Was going to do it once an hour, but will change it to 2-3 times a day and hope for the best. Maybe I don't know much, but 35 years working as a software engineer and developer, I may have picked up a few things. Actually never thought about SSD's failing sooner than spinning disk because of excessive writes until one of our hardware engineers brought this up as many of the SSDs that we sold a few years ago are starting to exceed their life expectancy and when you have maybe 1,000 of them in one appliance, that can be maintenance headache for sure. As for life expectancy of storage, got a WD Purple 3TB because that's the drive they recommend for NVRs. Well that lasted a week until it died, just got the warranty replacement yesterday.

is 8.0.2 safe going to? Just gun shy after 8.01.

Watching 26 1MP streams is going to use a heck of a lot of CPU power, so check that to see if that's the problem. You can use sub-streams if your cameras and NVR solution supports that. Also, if the different monitors will be displaying the same exact thing, the easiest thing to do is get an HDMI gizmo that broadcasts the same HDMI output to multiple monitors.