For my fourth year project, we are building a smart IoT traffic intersection, we need 4 IP cameras to use to detect cars and lanes, these cameras need to connect to our Cisco router and send the data there so the router can apply the image processing code locally. Where do you guys think I should start in figuiring out how to connect the IP camera to the router? and which cameras do you recommend? Our budget is 500 dollars (Canadian).
Which IP camera should I use for my fourth year project?
how to connect is easy: ethernet. this is an IP cam forum, after all. Wi-fi can be handy for certain uses,
but you still need provide power somehow...
if you don't already have one, you'll need some sort of networking switch (unless your networking setup already has that)
but used switches with PoE (Power over ethernet) to power the cams are available for $50 USD or less.
what sort of resolution, frame rates, and field of view (angle) will you need?
eg, you need to cover two lanes of traffic from 60ft? you need to ID license plates day & night?
etc. would really need more info to recommend specific cams...
but you still need provide power somehow...
if you don't already have one, you'll need some sort of networking switch (unless your networking setup already has that)
but used switches with PoE (Power over ethernet) to power the cams are available for $50 USD or less.
what sort of resolution, frame rates, and field of view (angle) will you need?
eg, you need to cover two lanes of traffic from 60ft? you need to ID license plates day & night?
etc. would really need more info to recommend specific cams...
So lets say I connected it using ethernet, will it have its own IP address? Because I am using openCV and to open a vid, it needs an IP address as a link.
As for the camera, 720p resolution is fine, frame rates 30fps is fine, angle is not important since it will only be setup to point in one direction, no need to move.
I would need to cover up to 5 lanes, the camera will be attached to the traffic light, how long do you think that is? and no need to ID license plates.
Thank you so much for the help, I really appreciate it
Q™
IPCT Contributor
I believe that you are underestimating the cost which IP camera equipment will have in order to put an effective system in place.
One of the first and most successful deployments of video cameras to detect vehicles in the roadway, process the image and provide vehicle detection input to the traffic signal controller is the Autoscope system. The Autoscope system was developed by the University of Minnesota and Image Sensing Systems, Inc. about 1990. Econolite of Anaheim, CA currently sells the systems.
I was the traffic signal and street lighting supervisor of a city in northern California and in 1993 administered the city's first installation of Autoscope at a signalized intersection. Being 25 years ago, the new technology was not without its hiccups but by 11 years later when I retired in 2004, many improvements had been made. I can only imagine that now,15 years later, that the system works remarkably well.
After spending my first 5 years beginning in '73 sawing the pavement with a water-cooled, diamond blade concrete saw and winding and sealing close to a hundred inductive loops in the roadway, I was eager to see better technology come along to detect cars. Inductive loops worked the best of all the devices that had come along up to that date but they left much to be desired: sawing a 2 inch deep slot in a 6 foot wide and 50 foot long rectangle in a perfectly good roadway begins the process of degradation of the pavement, not to mention the danger to installation personnel and the inconvenience to motorists while you are in the roadway installing them. They create ongoing maintenance issues.
My biggest question is this: will this project be a design that is to be theoretically functional in a full size signalized intersection or in a scaled-down intersection such as in a classroom or prototype environment?
I ask because the distance between the actual vehicle detection zone and the mounting location and height of the detection camera as well as the actual dimensions of the detection zone(s) will determine the lens specifications of the camera particularly focal length / field of view.
This will also determine the cost of your project, as @Q™ mentioned, you will likely have to increase your budget.
I suggest you read this as well: Using Machine Vision (Video Imaging) Technology To Collect Transportation Data
It would especially helpful if you've had too much caffeine and need help falling asleep.
Good luck on your project!
I was the traffic signal and street lighting supervisor of a city in northern California and in 1993 administered the city's first installation of Autoscope at a signalized intersection. Being 25 years ago, the new technology was not without its hiccups but by 11 years later when I retired in 2004, many improvements had been made. I can only imagine that now,15 years later, that the system works remarkably well.
After spending my first 5 years beginning in '73 sawing the pavement with a water-cooled, diamond blade concrete saw and winding and sealing close to a hundred inductive loops in the roadway, I was eager to see better technology come along to detect cars. Inductive loops worked the best of all the devices that had come along up to that date but they left much to be desired: sawing a 2 inch deep slot in a 6 foot wide and 50 foot long rectangle in a perfectly good roadway begins the process of degradation of the pavement, not to mention the danger to installation personnel and the inconvenience to motorists while you are in the roadway installing them. They create ongoing maintenance issues.
My biggest question is this: will this project be a design that is to be theoretically functional in a full size signalized intersection or in a scaled-down intersection such as in a classroom or prototype environment?
I ask because the distance between the actual vehicle detection zone and the mounting location and height of the detection camera as well as the actual dimensions of the detection zone(s) will determine the lens specifications of the camera particularly focal length / field of view.
This will also determine the cost of your project, as @Q™ mentioned, you will likely have to increase your budget.
I suggest you read this as well: Using Machine Vision (Video Imaging) Technology To Collect Transportation Data
It would especially helpful if you've had too much caffeine and need help falling asleep.
Good luck on your project!
Last edited:
By "traffic light" do you mean the actual signal head that looks like my avatar to your left?
And do you intend to actually install the camera(s) and equipment at an existing, in-use, signalized intersection?
Last edited:
Yes, most IP cams provide an RTSP stream (H.264 encoded) that your application can pull over the network and process.
5 lanes of traffic. hmm. so two lanes in each direction and maybe a left-turn lane in the middle.
Figure a lane is 8ft wide x 5 is 40ft width, just as a simple assumption.
From 40 ft away (other side of intersection), 40ft width give you ~53 degrees field of view (check my math?),
so common 4mm lens would be somewhat wider than you need. maybe pull back a bit and use 6mm from 50ft.
per IPVM calculator (here: IPVM Camera Calculator V3 ) , a 6mm lens gives you about 48ppf (pixels per foot)
resolution at 50feet with a 2MP cam (1080p). Is that going to be enuf? if not, use higher MP cams, but more $...
2MP cams can be found used for relatively cheap on Ebay or CraigsList, as people upgrade to newer high-er rez units. OR even new on Amazon within your budget, for example:
https://www.amazon.com//dp/B07JNC2Y2H
Stick with a name brand like Hikvision or Dahua (or one of their OEM'd versions.)
5 lanes of traffic. hmm. so two lanes in each direction and maybe a left-turn lane in the middle.
Figure a lane is 8ft wide x 5 is 40ft width, just as a simple assumption.
From 40 ft away (other side of intersection), 40ft width give you ~53 degrees field of view (check my math?),
so common 4mm lens would be somewhat wider than you need. maybe pull back a bit and use 6mm from 50ft.
per IPVM calculator (here: IPVM Camera Calculator V3 ) , a 6mm lens gives you about 48ppf (pixels per foot)
resolution at 50feet with a 2MP cam (1080p). Is that going to be enuf? if not, use higher MP cams, but more $...
2MP cams can be found used for relatively cheap on Ebay or CraigsList, as people upgrade to newer high-er rez units. OR even new on Amazon within your budget, for example:
https://www.amazon.com//dp/B07JNC2Y2H
Stick with a name brand like Hikvision or Dahua (or one of their OEM'd versions.)
As an Amazon Associate IPCamTalk earns from qualifying purchases.
Yes, similar to your avatar, we do not intedn to install them, we just need to connect them to the router and test them on a small scale model that we are building, but we need to mention that these cameras will work on a real-life scenario, and they should.
Thank you! This should do it! Really appreciate it ^_^
As an Amazon Associate IPCamTalk earns from qualifying purchases.