Average Frame Rate Video Surveillance 2016

Published Feb 11, 2016 05:00 AM

What is the average frame rated used in video surveillance systems?

Historically, 30fps has been considered 'full' frame rate and, in the past few years, 60fps video has become increasingly common.

Previous IPVM survey results found the average frame rate in 2011 was in the 6-8fps range. Now, more than 4 years later, how much has that changed? 150 integrators told us what frame rates they are using and why. Inside, we breakdown the results.

Statistics / Summary

***, *** ******* ***** **** ** ~10fps, ** ********** **** * ***** ago, *** ***** *** ***** **** almost *** ******* *** ******* ** delivering. *** ***** ***** ***** *** breakdown ** ********** *********:

*** **** ******* ***** *** *** steep **** ** *********** ********* **** or ****, ***** *** ** */**** since *** **** ******. **** ** less **** ***** **** **** ****** motion (*.*., *******) ** *** ********** frames *** *** ******* ********** ************* at * ***** **** (*.*., * face ******* ** *** ******). ** storage ***** **** ****** *****, ****** all *********** **** ***** **** **** such *** ***** **** *********.

*** '*******' ********** ***** **** ~**** to ~*****, * ***** **** *** ***** **** ***** *** ***** ************ ***** ** ****** ***** *** ********* surveillance *****.

Integrator *********

**** *** ****, *********** ********** **** ~10fps ******** * **** ******* ******* quality, ******* *********** *** ****:

  • "***** ** ** ****** *** ****** typically ***** *** ****** **** *** strikes * ******* ******* ********* *** retention **** *** ******* ***** **** enough ****** ** ********* *********** ** internal ******** *****."
  • "** ***, **** ******** ******* ******* requiring ******* ******* ******."
  • "**. ***** ** ** * **** compromise ******* ********* **** *** ****** video."
  • "*-** ***. ********* *** ********* **** only ******** **** *** *****. ** most *****, ** *** ***'* *** what *** **** ** *-** ***, you ***'* *** ** ** ** fps."
  • "*****. **** ***** ** ***** ****** and ******* *********/******* ************."
  • "** *** ***** ** ** *******. It's **** **** ****** ** *** a **** **** ** **** ******** (or ****'* *********), *** **'* *** a ***** ****** ** *******."
  • "** **** ***** *** ******* "****" video *** ******* ************"
  • "**-*****- ***** ********** ******* *** ********* storage ***** *** ******* *** ** network."
  • "* ***** *** *** ******* ** 10fps. ** ***** ** ** * good compromise between ***** ***** *** **** ****** to ******* *** ******."

**** ** ** ******* ** ***** vs ***** ** **** *** * person ******* *** ******* ***** **** - * ****** ******** *** ***** rates ***** *** ******** ****** ** movement:

**** ***** **** ***** ** **** than ****** ** ****** ****. *** more ** ***** ******** ** *** ***** **** ***** *** ***** ************.

The '*****' ***** ****

** ** *****, ** ** *** contend **** ***** ** '*****' *** everyone. ****** ~**% ** *********** ******** 15fps ** ****, ** ***** ** certainly * ******** ******** ** ************* who ** *** ************* ****** ***** rates. **** ** **** ** ****** by *** *********** (*.*., ***** ********) or *** ******** (*.*., ******* ********** applications **** ******* ***** **** ************).

** ** ******* **** **** ** less **** ** ********* ****** ***** *** relative ****** ***** ** ******* ***** ****** than **** *** *** ***** / issues ** ****** **** ******** ***** and **** ************* ** ******* * clear **** ** * ******* ** incident.

Resolution ** ***** ****

*************, ** *** **** * **** period, ** ******** **** ******* ********** has ********* ~***% **** ** / **** in **** ******* ** ****. ** ********, ********** (***** *****) has ********* ** ~*** **** ** frame ****, ******* *** ********** ** more ********** ** *** ** ***** surveillance.

Increases ** ******* ****

******* ** *** ******** ** **** resolution (***********) *** ***** **** (******), compared ** *** ******* ** ******* costs ****** **** **** *****, ***********, on *******, *** *** ****** **** per ******* *** ****** **** **** did ******* ** *** ******. *******, they **** ***** *** ********* ******* now ********* *** ********* **** ********* to **** ********** *** ********** ** frame ****.

Frame **** *******

** ******* * ****** ******** ** average ***** **** **** *** **** few *****. ** *** **** ****, storage **** ******** ** **** ** price *** '***** ******' ****** ****** bandwidth ******* *******. ** *** ******** ****, 10fps ** ******* **** ****** *** most ************ ******** ** ** ** not *** **** ** ******** ********** an ********. ** ***, ** ***** an ******** ** ******* ** **-***** in *** **** *** ***** ** likely.

Comments (28)
Avatar
Luis Carmona
Feb 11, 2016
Geutebruck USA • IPVMU Certified

How many integrators came across a situation where 5FPS, or even 10FPS, did not capture something important? Not counting special situations like high speed business processes or gambling or anything like that. I mean general use and overview.

(1)
JH
John Honovich
Feb 11, 2016
IPVM

We can't answer that precisely from our survey results. Extrapolating from the numbers and commentary provided, evidently very few have had issues at 10fps.

How much more of an issue / risk is 5fps than 10fps is harder to calculate but, at 5fps, video is definitely more jerky so that could be a factor as well. By contrast, 10fps looks fairly smooth to most people watching surveillance video.

Avatar
Luis Carmona
Feb 11, 2016
Geutebruck USA • IPVMU Certified

Sorry, I should have been more specific, I was actually wondering from other integrators in their experience have they ever had a system recording at 5 or 10 FPS if there was an incident where that one frame not taken in between two frames in a 200 millisecond interval was ever a problem?

UM
Undisclosed Manufacturer #1
Feb 11, 2016

I find it interesting how many people leave a camera at the default of 30fps, or ask the manufacturer "what do you recommend?"

JH
John Honovich
Feb 11, 2016
IPVM

What portion of your users do you believe leave the camera at 30fps?

UM
Undisclosed Manufacturer #1
Feb 11, 2016

I see a good 33% leave it at the default 30fps, another 33% put it to 15, and the other 33% at 7 or 10fps.

Just my experience.

(2)
(2)
Avatar
Jon Dillabaugh
Feb 11, 2016
Pro Focus LLC

What about the other 1%?

UI
Undisclosed Integrator #2
Feb 12, 2016

The other 1% for sure are confused and the not even recording at all, LOL.

(1)
(5)
Avatar
Jon Dillabaugh
Feb 12, 2016
Pro Focus LLC

But they hold over 50% of the worlds wealth? Surely they can afford a top notch integrator to enable recording!?!?

(6)
Avatar
Jon Dillabaugh
Feb 11, 2016
Pro Focus LLC

We became used to 7FPS due to limitations set by Dahua DVRs back in the SD analog days. When we began installing IP cameras, we were using the same Dahua NVRs, which had the same limitation. After finding clients that needed better recording systems (VMS), we still stay close to that rate, but usually default to 10FPS, unless there is a strong reason why the client needs more or less. I can't think of many that did need more. Usually it is for an identifying view like a door shot in a retail store. The added frames might yield a cleaner image. But that is usually just 15FPS at the most. We have never had a need for more than that.

(2)
(2)
RW
Rod Warren
Feb 12, 2016

Did your survey determine whether or not these frame rates are motion based recordings or continuous? Does anyone promote the concept of boost recording?

JH
John Honovich
Feb 12, 2016
IPVM

Rod, good question!

There was a notable minority of respondents who used boost recording, but it definitely was not typical.

In a future survey, we'll ask about motion vs continuous recording. Here's our old results on that: Surveillance Recording Mode Statistics

(2)
Avatar
Jon Dillabaugh
Feb 12, 2016
Pro Focus LLC

We used to try that with Dahua DVRs/NVRs, but you end up with fragmented files when trying to export an event. Our standard setup before realizing this was lower resolution/frame rate/high compression constant record with high res/frame rate/low compression on motion. Doing this on the Dahua recorders saves storage space, but is very difficult to export one seamless video when you have overlapping segments of non-motion and motion events.

Since drives are much larger and cheaper now, we simply keep the constant record and motion record setting the same, but leave motion enabled for the alerts/flags only. This allows you to export much easier, more homogenous video clips.

(3)
U
Undisclosed #3
Feb 20, 2016
IPVMU Certified

It sounds like Dahua is closing and re-opening a new stream with the camera at the new frame rate. I've also seen them glitch-up and miss a frame or two right when you want them to speedup.

But I'm not sure why they are written like that, since they could just have the camera stream at the high rate and discard frames (assuming it was a multiple like 5fps continuous to 15fps motion) when there was no motion.

This way it would never glitch, or create seperate chunks.

Of course, for those using boost trying to limit network traffic or camera/NVR CPU, this would not help. But for those caring only about storage there should be a checkbox for Server-side boost.

Maybe some have this?

JG
Jose Gerardo Rosiles
Feb 20, 2016
IPVMU Certified

I would expect that video analytics applications require a smooth video progression (i.e., higher frame rate) to be able to capture events and behaviors. Is there any study that shows the impact of low frame rates on VA performance?

Avatar
Ethan Ace
Feb 23, 2016

Analytics generally don't require a very high framerate, and some (camera-based especially) simply can't process very fast framerates.

A few platforms I know of give FPS recommendations in the 5-10 range. There may be some that look for more than that, but I'm not familiar with them.

If you go lower than 5 FPS, you do potentially lose a lot of motion information, with the object moving much more between frames than in higher framerates, so instead of the analytic being able to properly track its path, it sort of "jumps" from spot to spot, which may reduce accuracy.

(1)
TH
Truman HW
Jan 31, 2018

Do all brands define "motion" the same?

Is it a change in a certain number of pixels touching each other? Or is it cumulative, like the noise in dark scenes with amped gain?

Is a lower frame rate prone, for many cameras models (quality/cost level..) likely to increase the quality of the video as the processor, codec, etc., have a lower work load? 

I think I would take fewer images for higher quality images, for instance. 

 

(Another edifying thread. Danka)

JH
John Honovich
Jan 31, 2018
IPVM

Truman, motion detection algorithms vary by manufacturer. The most common approach is some variation of detecting the number of pixels changed, with a user configurable slider / option to adjust sensitivity / the total number of pixels changed and the areas where they are counted.

Lower frame almost never increases the quality of the video feed nor overloads the processor. There are processor constraints on cameras but to hit them typically one needs to run many streams or run applications and analytics, etc., not simply max out the frame rate.

JH
Jay Hobdy
Jan 31, 2018
IPVMU Certified

I believe we recently had a vendor adjust some of our cameras to 30FPS using h265.

 

We were using 15FPS with H264.

 

Thoughts?

JH
John Honovich
Jan 31, 2018
IPVM

adjust some of our cameras to 30FPS using h265.

We were using 15FPS with H264.

What camera was this? Why did they adjust it? It's possible on some newer cameras that support H.265 to do this.

One other question - why did they change it from 15fps to 30fps? Do you prefer that or?

UE
Undisclosed End User #4
Feb 09, 2018

WOW, I am amazed that people run 5FPS, quite frankly that is CRAZY in today's low cost storage world.  We run minimum 30 FPS and cameras that run 60 FPS I am utilizing it where it makes sense.  It may be the industry I am in (gaming) but I cant imagine a reason to run such a low frame rate with the low cost of storage.  I have 10 archiving servers + 2 Directory servers and with the archiver's at 14 day retention, they are still running all under 50% storage space consumed.  I am looking at cameras to push up with even further retention times where it may pay off. 

 

Since I have only worked in gaming, I guess I am kinda shocked there are people using such low frame rates.  

UM
Undisclosed Manufacturer #6
Feb 09, 2018

 It may be the industry I am in (gaming)

Since I have only worked in gaming, I guess I am kinda shocked there are people using such low frame rates.

With all due respects, this is the reason why it is understandably surprising for you. You work in gaming, and gaming has a lot of money it can spend. Gaming is your real world scenario, but there are other real world scenarios with other factors to consider.

(2)
TH
Truman HW
Feb 11, 2018

Basically - this is why manufacturers give you a range, instead of hardcoding it. :) Riiiiight? :) 

(1)
(1)
U
Undisclosed #3
Feb 09, 2018
IPVMU Certified

...but I cant imagine a reason to run such a low frame rate with the low cost of storage.

90 or 180 day retention is not unheard of.

UE
Undisclosed End User #4
Feb 09, 2018

Well I had a long reply but my new keyboard killed that.  So I was saying that I understand the trade off between lower frame-rate and storage, and 14 days my not be close to 90-180.  But frankly us having 90-180 days of storage would not benefit us at all.  After 14'ish days all it would do is encourage local LE to request old video and instigate a investigation on our end that is time consuming, and not beneficial to us.  Any in house employee theft is going to be found in a couple of days, too many checks and balances on money.  Unaccounted cash, IE: count is watched like no other activity in a casino. Dealers are watched very closely.  In gaming high frame-rate during reviews is key as we NEED to see small details.  I was mostly kinda shocked there were people using those small FPS rates, as I mentioned in my closed gaming environment that is unheard of, I was not saying it was wrong. Most of our claims to the casino are accidents, and having full frame-rate of some jackass running before falling, driving too fast in the parking garage, etc,  usually stops the plaintiffs attorney in their tracks.   

 

 

UM
Undisclosed Manufacturer #5
Feb 09, 2018

It's important to distinguish between live viewing and investigative playback. I haven't studied visual science and psychology but have read a lot about it. It is extremely complex. 

From what I've read, it seems a reasonable metric that the human eyes and brain cannot distinguish the difference in IPS from live video streams above approx 13-18 IPS. If viewing live video, there is no difference between 18IPS, 24IPS (movie theater), 30/60IPS (interlaced video), and 100/500IPS (machine vision). 

When investigating recorded video that dynamic changes quite a bit as the gaming investigators will tell/show you. High speed activity requires high speed recording to distinguish extremely slight changes from image to image (gaming slight of hand). This also applies to any high speed scene (traffic, machine vision, etc).

Equally important, as it pertains to video surveillance, is shutter speed. No matter how many IPS you record at a slow shutter speed, your high speed recorded evidence is likely blurred by slow shutter speed and rendered useless. 

AS
Ashley Schofield
Feb 11, 2018

If viewing live video, there is no difference between 18IPS, 24IPS (movie theater), 30/60IPS (interlaced video), and 100/500IPS (machine vision).

I would disagree with this, as when I was testing 2 camera's for capturing license plates, I had one @ 60fps and the other @ 25fps and you could tell that the 60 fps was a lot smoother...

quick eg below...

UM
Undisclosed Manufacturer #5
Feb 12, 2018

Live viewing or playback?