Flock Safety Falcon LPR Tested

By Derek Ward, Published Jan 06, 2021, 10:56am EST (Research)

*** ******* ******** ****** **** $** *******, ******** ***** ******* ***** ******* "stop *****", *** *** **** ** they **** ** *** **** *****?

IPVM Image

** ****** ***** ******'* ****** *** camera *** **** *** *****, ********* the *********:

  • *** ******** *** ******* ***** ***** overall?
  • *** **** **** ** ****** ************ common ******* ********** *** ********* ******?
  • *** ******** ** ** ***** **************?
  • *** ******** *** ******* **** *** type **************?
  • *** **** ** ****** ******** ****** in *** *****?
  • *** **** ** ****** ******* *****?
  • *** **** *** ***** ****** ***** interface *******?
  • *** **** **** *** ***** ***** work ** ***** *** ******?
  • **-****** ***** ****** ** *** ***** UI *** ******

Executive *******

** *** *******, ***** ****** ********* recognized *** **** ******* ****** ******, day ** *****, **** **** ********* between ~** *** (~** ***) ** to ~** *** (~*** ***) *** correctly **** ******* ********** *** ********* plates ***** **** **** ****** *** systems ** **** *****. ************, ***** did *** **** ***-***** ****, *******, or ****** ** ******, ***** ******* false ***** ** ***** *** ***** and *** ***** ** *** **** to ******** *** ***** ********, **** downloads, *** **** ********.

*******, *** ****** ********* ************* ******* plate ***** (*.*., ************ ********** ** Maryland), ******* **** (*.*., ********* **** for *****), *** **** (*.*., ********* Toyota *** ******). ************, *** ****** is ********* **** ***** ****, **** no ****** ** **** *****, ***** may ** ************ ** **** ************ with ***-******** **********.

***** ***** ***** *** *** *********. In *** *******, ***** ****** **** rarely **** ********* ****** *** *** and **** ** *****.

**** ** ****** *** *********.

Video ********

Flock ******* ******** *** *******

***** ******** **** ***** ***** ** involved ** ***** ******* *** ******** install, ***** ** **** *****, *** configuration. ***** ******* ******** ** ***-*****, who **** ********* **** *** ****** knowledge ** ********* *** *********, ****** some ***** *** ****** ** ****-******* may **** ** *********** (****** **** is ******** ** *** ****).

Accurate ***** ****** *****

**** *** ****** ** *+ ***** of *******, ***** ********** ********** **** license ****** ******* *** ******, **** accurate *********** ** ******** ******, ***** below.

IPVM Image

******** *** ******* ** *****, **** the ****** ***** ****** ** ************ when ******** **** ** ******** ********** plates.

IPVM Image

Two **** ********

***** ********* ***** ******** ***** ** "1.5 *****" ** *******, ******* * single ****** *** ***** *** *****, as ******* ****** *** ********* *** present ** *** ***** ** **** lane. **** **** **** ** ***** documentation, ********* *** ******'* ******** ****** two ***** **** *** ******* **** of **** ** ****:

IPVM Image

Non-Plate **** *******

***** ****** ******* ***-***** ****, **** as ********* ** *** ***** ** vans/trucks ** **** ** *******, ********, or ****** ***** **** *** **** issues *** **** ******* ** **** tests.

IPVM Image

Multiple ****** **** **************

**** **** ****** **** ******** ******* plates ** *** *****, ***** *** able ** ********* **** **** ** the ****** **********.

IPVM Image

Stacked ********** *** ********* ****** **** ********

******* ***** ********** **** **** ** Flock ****** **** *** ** ******, and ***** **** ** *****. ******* the ******* ********** **** ** *** left ** *****, **** **** **** correctly *** ******** **** *** ****** plate *******.

IPVM Image

No ******* ***** ***** / ***** ********

***** **** *** ******* * ***** to ********* ******* ***** **** ****** in *** ********, ********* ** ******* reads ***** ****** ** ***** **** captures. *** *******, ** *** **** below * **** **** ****** **** of *** *****, ********* ** **** four ********** ***** ****. ** *** right, ** **** ******* *** **********. Both ** ***** ****** *** ***** to * *********, *** ******* ** flagged *** ** ******* *****.

IPVM Image

Incorrect ***** ***** ******

***** ********** ******** ********* ******* ***** states, **** **** ****** **** ******* in ****. ** ***** ***** ***** was ********* ** ***** * ** 10 ***** ****** *** *******.

IPVM Image

Multiple **** ******

****** **** *****, ** ******* **** some ****** ***** ** **** ******** times, ***** ***** **** ** **** nuisance ******* ** **** ******* ****** a ***** ******. **** **** ********, it *** ****** ** *** *** vehicle *************** ******, ***** ***** **** searches ** ******* ****** *********.

IPVM Image

Incorrect ******* **********

****** *******, ** ******* ******** ********* vehicle ******, **** ** **** ****** SUV ***** ********** ** * "***".

IPVM Image

*******, *** **** *** ********* ********* or *** **********, **** ** *** Chevrolet *****:

IPVM Image

***** ****** **** **** ******** ** night, **** ******* ********** ***** *** recognized ** *** ** *** *****, only *** ***** **********:

IPVM Image

Front ***** ***** *** *********

***** **** *** ******* ***** ******* plate ***** ** ******, **********:

*** ****** ** ************ ******** *** engineered ** ***************** ** ********, ***** ** ********** across ******. **** **** ******* ***** plates *** *** *** ******** ** capture ** *** **** ****/********.

*** ******* ******** *****'* ********, **** extremely **** ******** ** ***** ******, even **** ******* ******* ****** *** day.

IPVM Image

US ****** ****

***** ****** ********* **** ***** **** US ******. ******** ****** **** *** accurately **** ** *** *****.

IPVM Image

Flock ****** ***** ********* ********

*** ******** ** *** ***** ******'* cloud *********** *******:

  • ***** ******: **** ****** ** ** *** cloud ***, ***** *** ********* **** the ******* ***** ****** ****, ********* the ********* ****** ******** **** *** be **** *** ******* ***/** *********.
  • ********* *************: ** **** ***** ****, ***** can ***** ** "******* **** ******" and *** ******* ******* **** ** color, **** *** ***** *** **** used ** ****** *******, ***** ***** be ****** *** ********* **** ** a ******** **** ******* ** * specific *****.
  • ****** *************: ***** ****** ******** * ***** notification ***-** ** *** ****** ** the ****** **** **** ****** ***** to *** ******* ** *** ***********, such ** ********* *** ***** ********.
  • **** ********: **** ******** *** ********** (***** Flock ****** ********** ** "*********") *** be ******* *** ****** ****** ********* or *** ****** ** * .*** file. ***** *** **** ******** ** .CSV *****.

*** ****** **** ****** *** *********** user ***** *** *** ******* (****** ~5 ******), **** ** ***** **** contains *********** **** ** ***** *********, the ****** ******, **** ******, *** the ****** *** *** ******* ***** is ** * *******.

IPVM Image

Watermark ****** *****

***** *** ***** ***** *** "*****" watermark ***** *********** ***** *** ******* plate ***** ****** * ******. ** asked ***** ***** ******** *** *********, and **** **** **'* ******** *** them ** ** ** ****** *** needed *** *************, ****** ***** *** no ******* *** ***** ** ** so.

IPVM Image

Physical ********

** *** ******** ********, ** **** a **** ** *** ******** ************, features, *** ************ ****** ** *** Flock ****** ****** *** ***** *****. As *** *** ***** ***** ******, power *** *** * *******, ** there *** ********** ******** ** *** test *****, *** ****** *** *******, there **** ** ****** **** *** camera ******* ***.

***** ******** ****-******* ************. ***** ********* why **** ** *** ********* **** mounting:

** ********* **** *** ***** ******* on * ****/**** ** ********, **** for ******** ******* (**** ** ******/********* modifications **** *****), *** **** ******* it **** **** ** ***** *** angle/vehicle **********.

*****

*** ***** ** * ******* ***** camera ****** *** "************" *** ** $2,000 *** ****** *** ****, ** explained:

******* ** ****** ********-*******. ****************, ** * ***-**** ********, **’*$*,*** / ****** / ****. **** ******** ********** – ************, maintenance, ******* *******, ******** *******, *** software *******.

*** "************" ****** **** ******** * single "********" ****, *** ** ***** configurable ********.

***** ** **** * "******-*****" ****** which ******** ******** *** *********** ** government ********, ***** ** ****** ** $2,500 *** ****.

*********-***** ****** ****** ** $*,*** / camera / ******** * ***-**** $*** ************ ****. This ******** – ***********, ******* *******, cellular *******, ********* ******* ******, *** software *******.

Comments (40)

Once again , Well done Ipvm

Agree: 2
Disagree
Informative: 1
Unhelpful
Funny

Great article one note this is a Nissan Xterra not a Jeep.

IPVM Image

Agree: 3
Disagree
Informative: 3
Unhelpful: 1
Funny: 1

I saw that too. I guess we can't be too hard on Flock misclassifying if IPVM also misclassifies. :)

Agree: 2
Disagree
Informative
Unhelpful
Funny: 9

Thank you Shawn and UD #1, that was changed!

Agree
Disagree
Informative: 1
Unhelpful
Funny

As a Jeep owner, I am outraged by this misclassification!

I wave at no Xterra's...

Agree
Disagree: 1
Informative
Unhelpful: 1
Funny: 12

Anyone able to say why they gave me an unhelpful when I was just trying to be helpful. I would hope that the IPVM team sees it as me just trying to help them maintain their high standards and not as me poking fun at them or trying to be difficult. If you can enlighten me on a better option for notifying them of something like this or simply share why you chose to vote unhelpful it would be appreciated.

Agree
Disagree
Informative: 2
Unhelpful
Funny: 1

#1, thanks for the help. I don't know why someone voted your comment unhelpful but I would not worry about it. I just gave you an informative vote to counteract that.

Agree
Disagree
Informative
Unhelpful
Funny

Thanks, there was no worry I only think I have paid for 1 renew in the last few years I was more just curious as to how I could improve my communication skills in the future or approach the suggestion of change better in the future. Don't usually see a lot of frowns, thumbs down yes, so was curious as to why. Thank you again.

Agree
Disagree
Informative
Unhelpful
Funny

Possibly a Jeep owner upset that they are now a part of Peugeot?

Agree
Disagree
Informative
Unhelpful
Funny: 2

Were there any temporary plates seen in testing, and how did the system do on those? Our experience in our market (TX) is that the vehicles we are asked to track the most have temporary plates. And since temp plates are paper and not reflective to IR, the same settings that read metal plates at night fail to read paper plates, and vice versa.

Agree
Disagree
Informative: 1
Unhelpful
Funny

There are other options on the market that will read temp and paper plates, even at night. Look at Rekor.

Agree
Disagree
Informative
Unhelpful
Funny

REDLPR is also really good at temp and Paper plates.

Agree
Disagree
Informative
Unhelpful
Funny

Hello UD #2,

During our time testing the system, there were only a handful of temp tags/plates seen, though during the daytime, they were identified without issue.

IPVM Image

As you mentioned, nighttime temp tags were less accurate, with low confidence reads common and dim/dark tags, as seen here:

IPVM Image

Agree
Disagree
Informative: 2
Unhelpful
Funny

Thank you for the detail. Your experience is what I would have guessed. We also see a lot of older vehicles with fake (home-printed), damaged/faded paper tags, and paper tags placed in plastic bags. The plastic bags (to keep them dry in the rain) are the worst, as the light will reflect off creases in the bag, making both day and night plates difficult.

Agree
Disagree
Informative: 2
Unhelpful
Funny

Wow, that's nuts! In terms of the plastics bags, I'm assuming they are the clear sandwich-esque baggies and not the grocery store plastic bag variety, because if that's the case, I don't know what LPR system would see those.

Thanks for the follow-up info!

Agree
Disagree
Informative
Unhelpful
Funny

Here is an example from earlier today:

IPVM ImageThe car dealerships will put the plates in these plastic sheaths that inevitably get crinkled or creased over time. And individuals making up their own plates (you can find a template for anything on the internet) will put them in Ziplocks or something similar. Now imagine either bright sun on the plastic, or it's night. People usually don't commit BMV's or robberies using their valid plate. Our hot list is mostly paper plates, as they will (fortunately) often re-use a paper plate in multiple crimes.

Agree
Disagree
Informative: 4
Unhelpful
Funny

What settings were used on the camera (shutter speed, iris, etc)? Does Flock allow the user to change them or is that hard coded in?

Agree
Disagree
Informative
Unhelpful
Funny

Hi Shawn,

Camera settings are hard coded, so there was no way for us to dig in to it and see what was being used. There are no camera settings to speak of Flock Safety cloud interface.

Agree
Disagree
Informative: 1
Unhelpful
Funny

Derek, Just curious. How you guys taken care of PII parameters during this testing. I could see so many plates you have captured during this LPR test

Agree
Disagree
Informative
Unhelpful
Funny: 1

guys taken care of PII parameters during this testing

In the US, I do not know of any legal issue in capturing license plates as we are doing.

If you are interested in this US approach on this topic and some of the more controversial practices, read this - This Company Built a Private Surveillance Network. We Tracked Someone With It

Agree
Disagree
Informative
Unhelpful
Funny

IPVM - well done as always. However, I did have a question about whether you did any testing with those smoke-colored plastic plate covers that are appearing more frequently. It looks like the plate shot above for "Flock Missing Vehicle Make Frequent Occurrence" has a clear plastic cover. I ask because we had an HOA that was telling us that they had been testing our installation of a DW tag camera for their neighborhood and we were consistently missing one of their resident's plate on his vehicle. After extensive testing, it turns out, this was a fairly new vehicle with a smoke-colored cover and the camera just wouldn't see through it at night. Once I looked it up, I also found that in GA, that sort of plate cover is illegal but I'm starting to see them so frequently now. If you get the chance, I'd be interested to hear how the Flock cameras perform with that sort of cover and are they able to discern the plate correctly.

Just as an FYI, the Georgia code reads:

No license plate shall be covered with any material unless the material is colorless and transparent. No apparatus that obstructs or hinders the clear display and legibility of a license plate shall be attached to the rear of any motor vehicle required to be registered in the state.

Agree
Disagree
Informative: 1
Unhelpful
Funny

I'll leave it to Derek to answer specifics on testing, but I only recently found out they're illegal in NY, as well, even clear ones. I put a clear one on my wife's car to try and keep some stuff off of it and she got pulled over and told to take it off.

Agree: 1
Disagree
Informative
Unhelpful
Funny

That is the funny thing with ALPR is that it is so easy to defeat. In the wintertime, there are lot more missed plates just from road dirt. Add some extra IR LEDs to the plate frame is another solution. Plate covers are another but they can be illegal.

IPVM Image

Agree
Disagree
Informative: 1
Unhelpful
Funny

Flock: The system is specifically designed and engineered to capture rear plates of vehicles, which is consistent across states. They will capture front plates but are not designed to capture at the same rate/accuracy.

IPVM: Our testing reflects Flock's response, with extremely poor accuracy on front plates, even when clearly visible during the day.

I read the first part of Flock's comment as 'every state has a back plate, but not all states have front plates, so we designed it to read back plates. Which makes complete sense.

What doesn't make sense to me is why, based on IPVM testing, does Flock perform so badly (comparatively) in reading front plates - if both front and back plates are identical?

Agree
Disagree
Informative
Unhelpful
Funny
Agree
Disagree
Informative
Unhelpful: 1
Funny

We are looking at bid where the basis of design appears to be Flock Safety and exploring other solutions such as AutoVu. & Vigilant Does anybody know if there is surcharge pricing for solar applications or have any solar reliability metrics?

The solar company I reached out to said that a 30W continues operational solution would be in the neighborhood of $4500.

Any feedback or advise would be greatly appreciated.

Thank you,

Agree
Disagree
Informative
Unhelpful
Funny

You should take a look at another company called Rekor.

Agree
Disagree
Informative
Unhelpful
Funny

This report does not mention the connectivity methods in the test setup. Was LTE/WiFi gateway used ? Does the camera cache the video frames/images? Or, direct streaming to the cloud app is required ?

Also, in the video for the setup there was a note regarding antenna port for wireless connectivity, but the tech noted "not in use". Is that WiFi or LTE connection?

Agree
Disagree
Informative
Unhelpful
Funny

Greetings Hamid, thank you for the questions!

LTE communications were used, not WiFi. Matter of fact, there is no WiFi connectivity with the Flock Safety device. Where did you read/see/hear that?

With regards to camera cache/direct upload, the camera uploads images immediately, so long as there is a connection. If the connection is disrupted/dropped, Flock said:

The amount of days of storage depends on how busy the road the camera is on is. On a neighborhood road it can be up to 14 days vs. 3-4 days for a busy surface road. The cameras will start to delete older data first and continue to capture new data.

The antenna port is for an LTE connection.

Agree
Disagree
Informative
Unhelpful
Funny

>> LTE communications were used,

What is the uplink requirements for the camera ? Mbps required to upload images ?

>> there is no WiFi connectivity with the Flock Safety device. Where did you >>read/see/hear that?

In the physical overview :

IPVM Image

Flock Safety Physical Overview

there was a reference to "Wireless Port" and noted it was not used. What is that about?

Also, is there any information of the camera teardown ? What SoC vendor? memory footprint of the camera ?

Agree
Disagree
Informative
Unhelpful
Funny

There is wifi connectivity. It is used during set up and then shut off.

Agree
Disagree
Informative
Unhelpful
Funny

Does anyone know if these cameras are NDAA compliant? I would think this is surely an issue if they are selling to government agencies and police departments without being compliant.

Agree
Disagree
Informative
Unhelpful
Funny

Flock replied saying their products are NDAA compliant. We did not do a teardown but we don't have any reason to believe this is not true.

Agree
Disagree
Informative: 1
Unhelpful
Funny

does Flock make their own cameras?

and if not, who does?

Agree
Disagree
Informative
Unhelpful
Funny

By the looks of this device, it appears that the Flock camera is just a single 5MP sensor which provides both plate capture and context. Is this correct?

If so, the LPR analytics are done at the server ("cloud")?

A comment about this test: It's one thing to verify how accurate the plates are read, it's another to see how many plates it never captured at all. This was tested over a period of two weeks, was there any analysis to see what it missed? This test seemed a little soft compared to other LPR tests we've seen here.

I have deployed various software LPR systems, and I've never found one that I would consider reliable for law enforcement. I suppose the nice thing about LPR, to Flocks advantage is, when you miss plates, no one is the wiser.

Agree: 1
Disagree
Informative: 1
Unhelpful
Funny: 1

Yes, the camera uses one sensor for overview and capture. Flock doesn't disclose exactly where all analytics are performed but mentioned in past discussions with them that LPR runs on the camera. If I had to guess [my opinion only], I would say that the camera is capturing metadata and sending it to the cloud for full processing. That's how a lot of metadata analytics function, e.g., Avigilon's H5A cameras capture metadata, but Appearance Search requires a server with a GPU for processing.

We didn't analyze for a full two weeks to see if it missed any vehicles, but we did look through a day's worth of video. We have cameras mounted out front recorded elsewhere and we verified that in a day it did not miss any vehicles passing.

Now, some caveats: First, our street is not extremely high traffic. I'd have to go back and look to see how many vehicles passed. It was probably 200+, but not anywhere near what a busy intersection might see. Second, speeds are relatively low. It's a 25mph speed limit zone, so speeds are generally +/- 10mph of that. Again, busier roads are likely to see faster speeds. Both of those issues are likely to increase chances of misses, but we did not see any.

Agree
Disagree
Informative: 2
Unhelpful
Funny

I was looking at a some information related to Flock Safety's Falcon ALPR. Is this essentially the same ALPR camera that was tested in this article?

Agree
Disagree
Informative
Unhelpful
Funny

Yes, #9, thanks. That's the product name of this, we did not list that correctly but it's now updated.

They also have announced a "Raven" for gunshot and audio detection, which we have covered but not yet tested.

Agree
Disagree
Informative
Unhelpful
Funny

Would be could comparison test with the new quick-deploy LPR from Motorola:

Agree
Disagree
Informative
Unhelpful
Funny

Mike, thanks, I had not seen this. We will do a post.

Agree
Disagree
Informative
Unhelpful
Funny
Subscribe to IPVM Research to read the full report.
Why do I need to subscribe?
The IPVM Research Service includes products tests and shootouts plus competitive and financial analysis, helping decision-makers better evaluate purchasing, partnering, developing, and/or competing against companies in physical security.
Already have an account?
Loading Related Reports