Hikvision LPR Camera Tested

By: IPVM Team, Published on Mar 26, 2018

Hikvision has expanded into the LPR market, claiming up to 99% capture rates, 95% recognition rates at speeds up to 75mph / 120kmh.

We bought and tested the Hikvision DS-2CD4A26FWD-IZHS8/P to see how it performs in real-world scenes and how accurate these accuracy claims are.

In this report, we examine:

  • Low speed/parking lot capture performance (5-15mph)
  • High speed/roadway capture performance (30+ mph)
  • Accuracy / performance
  • Capture issues (misreads, double reads, and missed captures)
  • Camera installation recommendations
  • Capture zone configuration
  • Plate search and export
  • VMS integration

And more. See our full results inside.

********* *** ******** **** the *** ******, ******** up ** **% ******* rates, **% *********** ***** at ****** ** ** 75mph / ******.

** ****** *** ****** the ***********-**********-*****/*** *** *** ** performs ** ****-***** ****** and *** ******** ***** accuracy ****** ***.

** **** ******, ** examine:

  • *** *****/******* *** ******* performance (*-*****)
  • **** *****/******* ******* *********** (30+ ***)
  • ******** / ***********
  • ******* ****** (********, ****** reads, *** ****** ********)
  • ****** ************ ***************
  • ******* **** *************
  • ***** ****** *** ******
  • *** ***********

*** ****. *** *** full ******* ******.

[***************]

*******

***** ** *** *****, Hikvision's **** *** ****** is * ****** ****** for *** ***** (*-*****) capture, **** ** ******* lots ** ******* *****, with ~**% ******** ** these ******, **** ** missed ******** *** **** few ********.

*******, ******** ******* ** ~80% ** ****** ****** (~35-45 ***), **** **** plates *** ******** ** all *** ******** ****** reading ******* *********, ******* only ******* ******, ********** characters, *** ******* ******** more **** ****.

************, *** **** *** limited *** **** *********** to *****-***** ***** *** no *** ** ****** plate ***** ****** ** the ****** **** *********'* own ****-**** (*** **** data ** ********* **** using ********* ****). ********** specifies ***/**** ***********, *** we **** *** *** tested **.

*******, **** **** *** 4A26/P ******** ****** ************* for ****** ******* ******** recommended ** *********. *** camera ******** ** ******* surveillance ******* ***** *** gain ******, **** *** turned ***.

*****

*** ********* **-**********-*****/* *** be *************** *** ~$***-*** ***. **** ***** ** comparable ** ******* **** cameras **** *** **** tests, **** ******************, ***** ** *** provide ***********. ** ****, for ***** ******* ***********, this ** * *********** discount ******** ** ********** separate *********** ******** ********.

**** **** **** ***** is *** ** ** confused **** *** ***-*** (i.e., *** /*)**** ******,***** **** ***~$*** **** ******.

Physical ********

*** **** ** ******* in ****, *****, *** features ** ***** ********* full-size *******, ***** *** same *******, *******, *** IR ************. ** ****** its ******** ******** ** this *****:

LPR ****** *****

********* ***** ******* *************** in ******** **************** ******, *****, ******* speed, ***., ********* ** the ******** ******** *** speed. ** *******, ***** are:

  • ******* *****: */**** ** faster
  • ******** ***** ** *********: 10-15°
  • ********** ***** ** *********: <30°, *** ** ******* as ********
  • ***: >** *** *** US ******, >*** *** wider ** ******.

**** **** ***** **** manually ****** ******* *****, enable ***, ****** *** capture ****, ***., **** installing *** ******. *** 4A26 ***** ********* ** 1/30s *** ******** **** typical *******, **** *** turned ***. ** ******** adjusted ******** ** ***** recommendations ** *** *****, changing ******** *** ***** of ********* ******** *****, discussed *****.

Fast ******** ****** ** ***** *******

** ************ ** ***** LPR *****, ******** ******* to */***** ** ****** results ** **** **** images ** *****, **** only ******* ****** *** lights ******* ** *** image:

***** ****** **** ** use * ****** ****** for ******* ******** ***/** color *********** ** ********.

North ******** **. ************* ******

**** **** **** ** a **** ** *** 4A26 *** ****** ** released ** ***** *******, on ***** ******** ***** license ******. ***** ** a **** ******* ******** in ****** ** *** US *** ****** **** in ****** *** ****, as **** ***** *** province ****** ******** ********, with *********** ** ******, contrast, ***. ** ********, plates ** *** ** and **** ***** ***** are **** ************ (***** x *****/**.*" * *.*"), with ****** ******** ****** and ****** **********:

******* ** ***** ***********, LPR *********** *** ** better ** ***** ***** areas.

Capture **** *****

***** ********* *** ******'* capture **** ** ********* the ****(*) ***** ******** will ****, ******* ** how ***** ***** ********* set ********* *****. ****, as *** *********'* *************, lines ****** *** ** strictly ***** ** ***** the ***** ** * road, *** ******* ******* where ****** *** **** in *** ***, ********** for ****** ******* ****** or ** *** **** of ****** ********, ******, or ********.

Low-Speed ******* (

** ****** ***-***** ******* in * ******* ***, with *** ****** ******* at ***** **** / 4m *** * ********** angle ** ********* ** ~10-15°, ***** *****. */***** max ******** *** **** in **** ***** ******* issue.

*********** *** ~**% ******** in **** ***** *** and *****, **** **** one ******** ** *** camera ********** * ********* to * ***** *** to *********** **** * chrome ******* ***** *****.

Poor **** ***** ******* ***********

*** ****** ***** ******* (35-45 ***), ** ****** on * ***-**** ****, with ********** ***** ** incidence ~**-**°, ***** *****. In **** *****, ******** dropped ** ~**%, **** many ****** *** ******** at *** *** ******** misreads *** ** ******** issues, ***** *****.

** *** *********'* ***************, we ****** **** */***** exposure ** **** *****, but ********* ****** ******** to */***** ** *** if ******* ********. *******, capture *** **** ******** was ******* *** *** night **** **** ***** exposure ********.

************, ** ********* ******** mounting ****** ****** *********'* recommendations, ******* *** ******** the ****** **** ~**' high ** ~*' ** a ******. ******** *** similar ********** ** ***** / ******.

Missed ********

** *** ***** ** road ******, *** ****** failed ** ******* ~**% of ****** (~**% ******* rate **. **% *********).

Misread ******

*** **** ****** ***** with ******* *** ********** simply ***** *******, **** other ******* ******** **** as **** ** ******* plate ****** ***** **** as **********. *** *******, in *** ***** *****, the ****** ******* * 1 ** *** ***** number *** ** *** license ***** *****, *** also ****** *** ***** character ** *** ***** completely.

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

******* ********** **** *********** read ** *** ********* on ******** ******, ***** below. **** **** ******* character *********** ** *** common, ********* **** ********* in ****-*** *** ******* used *** *******/*** *********** such ********* ******,*****, ***.

******* ********** **** ****** more ****** ** *** past *** ***** ** more ****** **** ********** additional ********* ******, **** some ******** ****** ****** for *************, ********* ********** ********. ********** *** ******** plates ***** *** ******* characters, ** ****. ***** plate ****** ******, *** generally ******** ** *-* characters ******** ***** *** spaced **** *** ********** stacked ** *** ****:

Partial ***** *****

** ***** ********* *** camera **** ********** **** of *** *****, ******* one ** **** **********. This *** **** ******** at *** ********* ** end ** *** ***** number, *** ************ ********** in *** ****** ** the ***** **** ******, as ****.

Vehicle ********* ****

****** ** ******** **** lettering **** ***** ******, with *** ****** ********* the *******/******* ** *** body ** *** ******* rather **** *** *****.

**** *** * ****** misread ** *** ****** as *****/******* ******** ****** by. ***** ** * collection ** * *** examples ** ********* *****.

Duplicate ********

*******, *** ****** ******** multiple ********* ********, ******* the **** *** ******** times, ***** ********** *** or **** *****.

Plate ****** ****** *** ******

***** *** ****** *** and ****** ****** ** specific ****** ***** *** web ********* ** *** 4A26. *******, ***** *** key ********* *********** ** this *********:

  • ***** ****** ****:**** **** **** ****** produces ***** ****** ****, not ***** ** *** plate **** ** ********. To ******** ***** **** the ******, ***** **** manually ****** *** ***** or ********** *********.
  • ** ******* ******:*** ***** ******* ****** does *** ******* *** preview ** *** *****. Users **** ******** *** image(s) ** **** **** images.
  • ******** ******** ****/*** ** Administrator:** ***** ** ******** images,***** **** *** ***** **** *** ** as ** *************. ****** otherwise **** ** ********.

Limited ***** ******* **** *******

********* ***** ******* **** support *** ******* ********* of *** *** ****** in *** ************** **** them ****** *******, ********* in ****** **** ***** or ***** ***** ** email ******* ** ****** contacts ***** ** *****.

**** ***** ***** ******** were ****, **** ******* gave ** ***** ** incomplete ***********. **** ***** if ** ***** ****** plate ****** ** ****-****, instead ** ***** *** camera ********, **** ********* that **, ** ***** not *** **** ** search *** ******, *** did *** ******* **** this ***** ** **** through ** *** ********* to ****. ** ****, despite ******** ********* ***** searching ****** ** ****, NVRs **** ***** *********.

Limited ****-**** ***********

***** *** **** **** iVMS-4200 *** ** ********* for *** *****, ** there *** ***** ********* LPR-related ******* (**** *******, License ***** *********, ******* Retrieval) ** ****-**** **** slightly ********* *********, *** none ** ***** *** installed ** *******.

**** *********** *** **** directly ** ****-****, *** "Road *******" ****** ** used ** ****** *** specific *****, ****** ******* lists, *** *********. *******, no ******* ***** ** displayed, ****** "******* ***** Retrieval", ********* *****.

NVR *********** ** ****

*******, **** *** ****** is ********* ** * Hikvision *** (****/**** -* models *********), *** "******* Plate *********" ****** ** used. **** ****** ******** read ****** ** **** as *** **** ***** of **** *** ******* archived *****, ***** *****.

*** ***** **** "******* Retrieval" ******** ******** *** supported ** *** *****, including ******* **** *** type *********.

HikCentral ********* "***" *******

**** **** ** ****** iVMS-4200 **** ***** *** vast ******** ** ********* systems *** . *********'* HikCentral ********* *********** ** LPR *******, *** ** have *** ****** *** capabilities. ****: ***/**** ** not ********* ** *** free ******* ** ********** available *** ******** **** Hikvision's ****(*). ************, ****/*********** * ******** ********* **********'* **** *******.

Third ***** *** ***********

*** **** **** *** been ********** ** **** VMSes, ********* ****** *** Mirasys. ** **** *** integrate **** ***** *********** to***** ********* ** *********, ********, and *****. ** *** positive ****, *****, *****, I/O ********** ******* ****** to ***** ***** ** our *****.

********

*** ********* ********, **** in **** ****, *** the **** ******* ******** available ** *** **** of *******.

  • **-*********-*****/*: **.*.********

Comments (67)

Great report IPVM on a decent LPR camera, but I gotta vent!

I do not like mathematical claims that include the term "up to". According to Hikvision this camera has:

• Capture Rate Up to 99%

• Recognition Rate Up to 95%

This weekend my Subaru's on-board trip computer was displaying up to 99.9MPG. Why doesn't Subaru market the new Outback as getting up to 99.9MPG? Marketing ethics?

Unfortunately 'up to' is used a lot in marketing. To your point, it is a lot more accurate to market 'normal' or 'average' use.

In fairness to Hikvision here, the numbers themselves tested true but only for slower speeds so if they would have limited the claims to that speed range, it would be appropriate.

Any video at night?

Yes, here's an example from the road at night, using 1/1000s exposure:

The only things visible are the plate and taillights: 

In fairness, isn't that the point of a LPR camera? If I were having one of these installed I would expect that it would be only for catching plates and if it did that job properly, I would be satisfied. Specialized equipment to perform specialized tasks.

I personally agree, and Hikvision does not market the camera as capable of both overview and capture. Some others do make that claim, but it generally requires two cameras for proper use.

I find that capturing plates in complete darkness is easier then night but with some light around. One reason is some cameras don't want to kick into black and white as well as other factors. Was this tested at night but with some other typical parking lot lighting on?

At the road shown in the report, it's dark. I don't remember off the top of my head, but there is little lighting there. I'd say maybe 0.5lx, tops.

In the scene shown for slower speeds in the parking lots, it's closer to 1-2lx.

The camera was definitely in monochrome, though, especially with exposure set to 1/1000s as recommended.

Do you think this is better, worse, or equal to Mobotix?  Also, it doesn't appear this vehicle is sitting still.

The nice thing about a solution like this is that it is an all-in-one deal and I imagine configuration isn't too big of a hassle.

With regards to Mobotix, I really like the M15/16 because you get your overview camera and tag capture sensor in one form factor. Set-up would be trickier for someone less experienced because everything needs to be set-up manually (including the IR unit). Apparently Mobotix is looking to do an onboard LPR application at some point, but who knows when that will be ready for market. Not everyone subscribes to the Mobotix way of doing things but I really like it for a litany of reasons. They are tough as nails and do not fall victim to the cyber security woes we so often hear about in these forums.

The nice thing about a solution like this is that it is an all-in-one deal and I imagine configuration isn't too big of a hassle.

One thing that was not clear enough in the report: you have to manually change settings to Hikvision's recommendations. The camera defaults to typical surveillance settings (1/30s max exposure, LPR analytic turned off). We manually adjusted these settings during testing.

Might have missed this but where does one find Hikivision's recommendations. Is it deep in the manual?

We had to ask tech support in the US, but they're pretty well documented in Europe and the rest of the world in this guide.

Well said.  Maybe IPVM will test the M15/16 for license plate recognition/capture.  They may not be as experienced as you, but seem fairly capable.  Green beer on the house next time!

Report like this, is the main reason of keeping me here for years.  Good work.

Why not an Axis or Vivotek test?
Enough of Dahua and Hikvision ... nothing works and you continue to emphasize the Chinese.

We tested this camera because it's Hikvision's first foray into an all in one LPR camera in North America, and it's at a much lower price point than typical systems.

As far as Axis and Vivotek, while they have cameras which can be used for license plate capture, they don't have full blown LPR systems, so what would you suggest we test? 

Note, we have tested Axis cameras with two LPR applications: Genetec Cloud LPR and ipConfigure.

I do not think Axis or Vivotek or another manufacturer, invest in your own OCR, and will not do it. Both companies are keen to partner with VMS, and are increasingly approaching them. But its cameras are excellent, deliver what is in the datasheet, has high durability and cherish for partners bins. I would test AXIS and Vivotek with Genetec and ISS and would also include Hanwha, which in my view, are the only 3 manufacturers still struggling against the Chinese and their products. Apart from Pelco, which fell by the wayside, which is a shame, I would focus on these 3 manufacturers. The rest, it's all OEM from Dahua and Hik. Test the VMS with these manufacturers.

We test Axis and Hanwha regularly, e.g., Axis last week and Hanwha last month and many tests overall in the last year.

As for Vivotek, we have not tested them as much. At the very least, they are less visible in North America recently and marketing has been lower. On the other hand, they are one of the larger independents remaining.

We are open to testing Vivotek so if anyone has good recommendations on what to test (i.e., things they think are novel or especially competitive) we'll consider it.

Related, as for Hikvision we are right now testing the DarkfighterX because the IR / color PTZ combo is novel (at least how it's described on paper) so expect that soon.

Dear John, I have signed IPVM for over 7 years and I really like the publications.
But the subject here is OCR. As I said above, it would be interesting to test ISS / Genetec with Axis, Vivotek and Hanwha.
I would like you to test the IP 816 LPC KIT (new model is on the way), we have had great success with this product.
And again, congratulations on the publications and tests.

I would like you to test the IP 816 LPC KIT (new model is on the way)

I think you are referring to this one - VIVOTEK's IP816A-LPC-v2 Kit

From the video, it appears they are expressly targeting this at high-speed applications, which are certainly harder as this Hikvision test shows. One thing I am not clear is that the video shows recognizing plates. So does this camera include the OCR / LPR software or?

John, the LPR / ORC can be shipped on the camera, but that was not the case. We have had a lot of success, especially at night, using this model with Genetec, ISS and Digifort. Remembering that this model, has unique characteristics, developed by Vivotek, I have not seen until today, no camera with a rating index so act at night .. Whatever you need, we can send it to you. J you mentioned cameras with new technologies, test line 65 (with anti virus). 

I know these are HD and the instructions stated a specific minimum pixel size that you tested to.

All marketing aside, I recall trying to limit LPR to a single one of traffic for each camera so these appear to be a wider FOV than I expected. 

How did it do with a FOV of a single lane?

How did it do at maximum distance assuming it’s varifocal and IR illuminated?

Great Comparison , but with out at least 4 others to compare to , only singles out HV

need multiple companys for a real comparison 

Like to see DW , Flir, KT&C, Axis, Arecont, many others compared to see in report what they added upto 

Also would like to see , speeds of 25,50.75 , day vs Night , and relationship of the day/nite shots 

Like to see many more different ideals 

Rain, Snow, Bright , Pitch Dark , Fog in some of the test to see how they stack up. 

 

What I don’t understand:

Why are the camera manufacturers trying to do LPR with standard CCTV cameras? LPR is already succesfully deployed in the market with dedicated cameras for over 15 years now, by companies like; Tattile, Vigilate, ARVOO, PIPS, JPL, Quercus, ARH and a couple of others.

Applications like Parking Access, Traffic Data Collection and Enforcement. However, the mentioned companies are not the usuals suspects in CCTV. Plates are read with consistant accuracy performance, day and night, because the plates are read with global shutter, synchronized IR flash and fixed IR filter. Standard output is a JPEG picture and LPR data string plus optional metadata. For overview an optional second color sensor is used in the same housing that sends out the H.264 stream.

Yes, these dedicated LPR cameras are a bit more expensive (but not that much) and they actually do the job properly.

Just offering CCTV based LPR cameras that will work but not rock-solid, at a lower price point into a relatively limited market size does not sound like a strategy that makes a lot of sense for the large camera manufacturers.

Anyone in the group that can share an opinion on this?

 

 

Is this not Hikvision's approach in general? It's a lot less expensive than almost all those companies you mention (Tattile, Vigilate, ARVOO, PIPS, JPL, Quercus, ARH) who tend to be 5x the price of the Hikvision model.

I think there's a lot of users out there who want to add LPR for simpler applications who could never justify the price before who will go for this.

I do agree with you that some significant portion of potential buyers are going to be burned by the accuracy issues the camera presents especially in more challenging environments.

 

Interesting, but it is an old way of doing LPR, OCR, ie using inductive loops, image in CIF, and as you mentioned in Jpeg (purpose and only pick up the plates and not the vehicles in movement as we currently see). On the test with HIK, I see that it is a basic situation, where any camera with a good WDR can do it (not a camera made for LPR), as placed. And how will you have the image being that the lens is focused on the board in H264? I did not understand the information I quote this model from Vivotek, because we tested it and we have more than 100 cameras fitted with this product, giving a 100% daytime and nighttime beating between 97 and 99%, which is very high in the night mode. Remembering that this model gives us the 2 way plates.

"Finally, the 4A26 has no plate data integration to third-party VMSes"

We have made Ksenos VMS integration for this camera, to open gates and other low speed solutions.

Mirasys do also have integration for HIKVisions LPR cameras. Blacklist/whitelist, open gates, save statistics etc.  

Pekka, thanks. Do you integrate the license plate data from the Hikvision camera, i.e., do you display the plate information in real time or allow searching for specific numbers? We'll update the post but want to clarify what specific integration you support.

John, the plate data is integrated to Ksenos database that allows user to search numbers and it works as access control system where plate is like a tag.  

While LPR is still pretty high end application, we are seeing more and more demand for license plate capture( where the plates are recorded and manually searched off the video). Can you do a License Plate Capture shootout? Many manufacturers claim that their cameras can do license plate capture but to what extent is the real question. 

Peter, we have done some in the past, e.g. 2016 License Plate Capture 4K Test, 2014 Low Cost License Plate Capture Shootout and License Plate Capture Shootout 2014.

We are definitely open to doing more but those 3 should help start give a feel about tradeoffs involved.

At the Hikvision Roadshow last week, they showed us a slightly tighter integration with IVMS-4200 their NVR. You could search by tag criteria, such as a partial plate, or the full tag if known. However, they couldn't figure out how to see a list (or export one) that showed a log of plates read. The underlying ANPR data was there, but it was only available via search.

Hi Jon, for the .CSV file you need must use the NVR hardware front end > VCA > Plate Search > select camera and date range > Search > deselect Picture and Record images leaving just backup license plate info > Export (to USB stick).

At present only available through the front though....no way in IVMS or browser.

The CSV file comes out clean as a whistle....

Update: Based on feedback from users, we have added a brief section showing integration between the 9000 series NVR and LPR camera attached, which does allow LPR search, event images, and video retrieval from iVMS-4200. As confirmed by tech support, there is no direct integration to the camera, but tied to an NVR it works as expected.

I just ordered this camera last week. Hopefully, will be able to test it with Hikcentral next week.

For your information, LPR is als supported by the less expensive 7700 series and there is a plugin available for Milestone.

Thanks, I actually checked the 7000 series and updated the report yesterday to be more clear. 

As for the Milestone plugin, where does one get that? I don't see any mention of it on Hikvision's partner integration status, nor do I see it on Milestone's partner add-on list.

Good article, Ethan but can I just add a few points?

We've used the LP camera across a range of Hiks NVR's, but as the LPR's are usually on higher end jobs - we settle on the 96 series more often than not.

The IVMS allows multiple LPR cameras to be searched simultaneously on the same or different sites - a really useful feature.

Not sure if I missed it in the write up - but bear in mind it can also support up-to 4 lanes of traffic.

Coming onto settings we found that 1/1000 exposure is best and gain set to 20, use manual focus and set day/night switch to "Triggered by video" with a sensitivity of 7. Smart IR "Off" of course.

There are annoyances for me - I'd like the camera set-up preferences to be consistent across the camera menu, NVR menu and IVMS, but we just work on the camera menu to have the full access to what we need.

In regard to misreads, we have occasional ones but what the article doesn't show is our anecdotal experience that if it captures text on a back of a vehicle in error, invariably it captures the VRN immediately afterwards as shown on the image below. (Notice the consistency of style on UK plates - white is a front read and yellow being a rear read).

4 Lanes of traffic? At night? That is very wide!

We recently put in about 7 of these cameras for a small town and they are very pleased with them.  I would echo Tim's comments exactly. 

Cameras do a great job, but not 100 percent accurate. 

Very pleased with the 96xx series NVR and 4.0 firmware for this application.

Occasionally picks up other text on the back of a vehicle: bumper sticker, vehicle model, sign graphics, etc.

Oddly will occasionally miss a slower vehicle, but will hit on a faster vehicle.

Definitely recommend putting microSD cards in all of the cameras as a backup.  The internal camera browser interface also had decent capture recovery

Which lenses was tested? Because for this scenario Hikvision recommend 8~32mm

If you are referring to my post Lucas, we only ever use the 8-32 lens. Closing down the lens causes issues with capture so it set, on this application, for the optimum FOV.

 

We tested with the 8-32mm lens. I find that normal varifocal lenses (2.8-12mm) are too wide for most applications.

NVR is not required for integration with iVMS-4200.

There are 2 modules in iVMS-4200, 'License Plate Retrieval' that works with NVR and 'Road Traffic' that can search plates stored in microSD in LPR camera and export image captures and data to an Excel file.

Both of them are pretty much the same, one for accessing NVR and other for IPC, I don't really know why Hikvision includes two modules so similar, probably they come from different R&D develops.

Hi, thanks for that info. What version of iVMS-4200 are you using? I have all the modules available loaded and do not see a road traffic option.

Also, Hikvision's USA tech support is apparently unaware of this, as well, because they told us iVMS could not search the camera.

UPDATE: Using the latest iVMS-4200 client we could find, 2.7.0.10 found on the European tech portal, plates are searchable and exportable directly from the camera in iVMS using the "Road Traffic" module. 

A full field of view capture from each read is available, but no linked video.

Ethan, using V2.7.0.10 it's lurking in the modules customisation link under the statistics heading. Having tried it - on my set-up it doesn't actually connect, whilst the LPR works fine - so I'd need to check that out. I can't see a .csv export either, but without data I don't know for sure:

 

 

Thanks! That's not in my 2.6 install from NA. I just found 2.7.0.10 on the Europe FTP and will check it out.

I'll let Hikvision North America tech support know, too.

I am assuming the LPR camera has to be installed low on the ground.  How do you protect the camera from vandals in a bad neighborhood?

Really the main issue is angle, not height. Many people mount them low so that they can minimize the angle of incidence, but you could also move the camera further back and mount it higher to view the same area.

Hikvision recommends the camera be mounted at an angle of no more than 30° to the ground. I'd personally try to keep it at 20-25°, just based on what I've seen with various LPR.

Has anyone had troubles with this camera telling the difference between zeros and the letter O? On some plates the camera also says the letter O instead of the number zero. 

Any recommendations? 

That is a common limitation to most LPR OCR systems, because O 0 Q, 5 S, 1 I, 2 Z, B 8, etc are OCR equivalent characters. Some systems have the option to enable flexibile or "fuzzy" matching which will still alert on a hit of a plate with an O instead of an 0. The Hikvision LPR User Manual does not list any official character equivalancy or flexible matching options.

If you are creating a specific black/white/hotlist you would need to enter in the combinations of plates possible (i.e ABC1234, A8C1234 A8CI234, etc)

So which is the best camera that can solve these problems?

based on my private test:

ARH's LPR software provides a tiplist for each character, as long as the input image quality fits the minimum OCR requirements.

Also ARH's SMARTCAM is able to differentiate the country&state (Type ID, not just text result), based on that syntax info, they are able to say whether plate characters are a letter (O) or a number (0 - zero). 

There is a Milestone Plugin for this that has its own Tab in Milestone. Works Brilliant and as is on the edge, No hefty LPR server.

Contact me for more details about the Smart Protect Plugin. 3rd Party not from Milestone and attaches to the SLC.

Thanks

Dan

I'd be interested in the Milestone plugin Dan.  Message me a link if you can.

It's been a little while since I visited this discussion, but just to add that we've found the end users to be more than satisfied with the performance. We manage expectations as t what may affect the accuracy of a read and the expected success rate. What some may overlook is that even a partial read is more than useful. By capturing the contextual video, a partial OCR read can be "manually" enhanced with the visual capture, which increases the read rate significantly (obviously doesn't help in data export, but does help in operational situations).

Whats impressed us is the ease of set-up, having used other systems in the past which were far more sensitive to height, angle, object distance than the Hikvision - which invariably burned cash in return visits and irate customers. 

I still feel this is a greatly overlooked product given the massive price difference to mainstream solutions and brings ANPR affordability to the majority of users with minimal engineering difficulty.

 

What some may overlook is that even a partial read is more than useful

While I agree with your general sentiment about its affordability, the 'partial read' claim is problematic. Many / most LPR users want to make automated decisions, send me an alert for "xyz123" or show me all the times that "xyz123" was captured. A 'partial read' (a euphemism for at least one character wrong returns errors in those common contexts).

To be honest - not in my experience John. The customers we've sold to, don't generally use t for automation but do use system as an extension of the security functionality rather than automation of devices. Typically this will be a distribution depot, large office car park, hospitals and universities who are using the system to check quickly if a certain VRN has entered (left) site, or to check suspicious activities. In the case of the hospitals with multiple car-parks its proved priceless in assisting patients to locate cars as the system will check all ANPR cameras simultaneously and indicate which car park the car is in. This is easy using a partial VRN entry.

But I do agree that where additional integration is required, this may be an issue. Let's also be mindful that in the UK the partial read issue is significantly less than the US as only one, very prescriptive, license plate design is used as opposed to the US where seemingly anything goes.

I think if you sat your customers down and showed them the costs associated with other systems including licenses, and the showed them the Hik solution with only marginally less accuracy but massively lower costs - the decision is easy. The fact it sits on a standard NVR and uses the common GUI just added to the value.

Horses for courses....

We installed 2 of them and found we needed to use IVMS-5200 to have good reporting of the licence plates. We still felt the accuracy is not perfect (80%) although acceptable for the small application. If it was a very busy entrance and exit I would say hivkision ANPR might struggle with the accuracy. 

We also tested dahua and found the results identical to hikvision. 

That's way below our experience of accuracy. There are of course so many parameters that it would be hard to compare. But we consistently had over 90% and more typically 95% on a single camera, 2 lane entry/exit from a car park reading UK VRN's and traffic speed was relatively slow. We closed down the FOV to a nice tight pinch point (32mm lens) and monitored the system for every vehicle read for several hours to get it right.

We also used the settings that I covered earlier in this article that is not the same as what the cameras come out of the box with but made a massive difference (thanks as ever to Dave at DVS for the spot on info).

 

I am awaiting delivery for one of these 4A26/P (8-32mm lens) cams. Thank you for the donation Sam!

Has anyone noticed significant improvements in performance with current recent firmware updates?

I want to try supplementary IR illuminators with this cam. Any Raytec models someone would recommend?

USA Vehicles ill be capturing will be 75-150 feet distance, at 15-30 MPH.

I am awaiting delivery for one of these 4A26/P (8-32mm lens) cams. Thank you for the donation Sam!

Has anyone noticed significant improvements in performance with current recent firmware updates?

I want to try supplementary IR illuminators with this cam. Any Raytec models someone would recommend?

USA Vehicles I'll be capturing will be 75-150 feet distance, at 10-30 MPH.

Nick, I checked the Hikvision website and it doesn't look like they updated firmware since we tested this, I emailed Hikvision to confirm if there is newer firmware and if there are performance improvements, I will report back when they respond.

I have FW V5.4.5 build 190402 for DS-2CD4A26FWD-IZHS8/P

Please PM me if you want to test it. I found the camera more stable and detection rate is higher with that version. I'm pretty sure it's not on their website (idk why), and it was sent to me by tech support after my complains

Read this IPVM report for free.

This article is part of IPVM's 6,435 reports, 865 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.

Already a member? Login here | Join now

Related Reports

Vivotek LPR Camera Tested on Apr 15, 2020
Vivotek has historically sold license plate capture cameras but not LPR. Now,...
Leica Launches LIDAR / Thermal / IP Camera on Mar 04, 2020
Swiss manufacturer Leica is launching what it calls a "real-time reality...
Hikvision Fever Screening Thermal Solutions Examined on Apr 13, 2020
Hikvision is marketing "safer, faster, smarter" with their Fever Screening...
Tiandy Super Starlight Cameras Tested on Apr 30, 2020
Tiandy is taking aim at China competitors Dahua and Hikvision, with a new...
Hikvision MinMoe Temperature Screening Terminal Tested on Jun 23, 2020
Hikvision has expanded its temperature screening options. We tested...
Seek Scan Thermal Temperature Screening System Tested on May 28, 2020
Now that IPVM has tested Dahua, Hikvision, and Sunell, we are returning to...
Euklis Presents AI Analytics on May 05, 2020
Euklis presented its AI facial recognition, LPR, and object recognition...
ZKTeco Body Temperature and Mask Detection Reader Tested on May 26, 2020
While dedicated fever cameras emerged first, now tablet/kiosk fever detectors...
TVT / InVid White Light Camera Tested Vs Hikvision ColorVu on Mar 18, 2020
With mega China manufacturers Dahua and Hikvision facing both bans and human...
Anyvision Presents AI Facial Recognition and Mask Detection on Jun 08, 2020
AnyVision presented its AI facial recognition and mask detection at the May...
HLC Camera Shootout - Avigilon, Axis, Dahua, Hanwha, Hikvision, Panasonic, Uniview, Vivotek on Mar 02, 2020
Highlight Compensation (HLC) claims to reduce the effects of strong light...
TVT Temperature Measurement Terminal Tested on Jul 23, 2020
While Dahua and Hikvision get the most attention for China temp products,...
Multi-Imagers Added to IPVM Camera Calculator on Jun 15, 2020
IPVM has added multi-imager cameras to the Camera Calculator, supporting this...
New Axis M30 Cameras Tested on Mar 26, 2020
Axis has released a new generation of, for them, relatively low cost M30...
IronYun AI Analytics Tested on Feb 17, 2020
Taiwan / US startup IronYun has raised tens of millions for its "mission to...

Recent Reports

VSaaS Will Hurt Integrators on Aug 06, 2020
VSaaS will hurt integrators, there is no question about that. How much...
Dogs For Coronavirus Screening Examined on Aug 06, 2020
While thermal temperature screening is the surveillance industry's most...
ADT Slides Back, Disappointing Results, Poor Commercial Performance on Aug 06, 2020
While ADT had an incredible start to the week, driven by the Google...
AHJ / Authority Having Jurisdiction Tutorial on Aug 06, 2020
One of the most powerful yet often underappreciated characters in all of the...
SIA Coaches Sellers on NDAA 889B Blacklist Workarounds on Aug 05, 2020
Last month SIA demanded that NDAA 899B "must be delayed". Now that they have...
ADI Returns To Growth, Back To 'Pre-COVID Levels' on Aug 05, 2020
While ADI was hit hard in April, with revenue declining 21%, the company's...
Exposing Fever Tablet Suppliers and 40+ Relabelers on Aug 05, 2020
IPVM has found 40+ USA and EU companies relabeling fever tablets designed,...
Indian Government Restricts PRC Manufacturers From Public Projects on Aug 04, 2020
In a move that mirrors the U.S. government’s ban on Dahua and Hikvision...
Directory of 201 "Fever" Camera Suppliers on Aug 04, 2020
This directory provides a list of "Fever" scanning thermal camera providers...
Face Masks Increase Face Recognition Errors Says NIST on Aug 04, 2020
COVID-19 has led to widespread facemask use, which as IPVM testing has shown...
Dahua Loses Australian Medical Device Approval on Aug 04, 2020
Dahua has cancelled its medical device registration after "discussions" with...
Google Invests in ADT, ADT Stock Soars on Aug 03, 2020
Google has announced a $450 million investment in the Florida-based security...
US Startup Fever Inspect Examined on Aug 03, 2020
Undoubtedly late to fever cameras, this US company, Fever Inspect, led by a...
Motorola Solutions Acquires Pelco on Aug 03, 2020
Motorola Solutions has acquired Pelco, pledging to bring blue back and make...
False: Verkada: "If You Want To Remote View Your Cameras You Need To Punch Holes In Your Firewall" on Jul 31, 2020
Verkada falsely declared to “3,000+ customers”, “300 school districts”, and...