Jump to content
Guest JOINDVR

Pixels per feet

Recommended Posts

Guest

Hi!

 

What is your experience regarding pixels per feet/meter for good facial video/images?

 

 

JD

Share this post


Link to post
Share on other sites

According to Arecont 100-120 PPF for facial recognition but they recommend you use the specs analytic software you are using. I haven't used facial recognition so I can't recommend from experience.

 

Arecont recommends 45PPF for LPR and I have had great results when I spec systems using this as a guideline.

Share this post


Link to post
Share on other sites
Hi!

 

What is your experience regarding pixels per feet/meter for good facial video/images?

 

 

JD

 

Hi JD,

 

I put it into 2 seperate sections, 1. Recognition, 2. Identification.

 

Recognition where the person is known or can be recognised by someone by the recorded footage.

 

Identification where the person is not known and requires comparison of the recorded footage to a person or other images in detail.

 

For Recoginition I try for the person to be clear with face ocupying 30-50% of vertical screen size. Generally in the 75-100 ppf range.

 

For identification with the face to be clear ocupying 100% of the vertical screen. Minimum of 150ppf but try to get 200+ ppf to get minor details clear, like scars and markings, eye colour, etc..

 

While giving an approx PPF, other considerations need to be taken into account, level of compression (always use the recorded footage as your reference if for example your camera dual streams), lighting etc..

 

I have some installations where the choke point cameras are giving 250+ PPF for identification, and they do their job well with a lossles compression.

Share this post


Link to post
Share on other sites
3. Overview 99% of all cameras !

 

Even with the overview cameras, if you've got the FOV set up so that the widest scene width still gives the correct PPF for good facial images then all good. Issue is that most don't, which means when you zoom in on a face off the recorded footage all you see is MR. Fuzzy.

 

Problem is most "punters" want the result but dont want to pay for the number of or quality of cameras to acheive it. (ok thats my whinge for the day).

Share this post


Link to post
Share on other sites
Problem is most "punters" want the result but dont want to pay for the number of or quality of cameras to acheive it. (ok thats my whinge for the day).

 

I think that's the "career whinge" for just about everyone here!

Share this post


Link to post
Share on other sites

yep thats what I mean, "how come i cant get the rego off the refection of a spoon" they do it all the time on TV....

 

These cheapo push-em out hard CCTV companies that throw domes about high on the roof...And a job well done ...here's the cash . aghh!

 

After the 10th robbery with the 'hat people' they usually get smart and get a pro in ! [who finally aims a camera at the door to get an ID]

 

z

Share this post


Link to post
Share on other sites
125532_1.png

 

This picture is quite misleading: no camera can output the picture (even the camera with 3CCD) as sharp as this one (because it is resized from a larger one).

 

In order to get similar result, you may need multiply the number of ppf by 1.3 or even 1.5

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×