Review: EmpireTech IPC-T58IR-ZE-S3 / Dahua 5842T-ZE-S3 - 8MP / 4K Turret Camera

[Wildcat_1]

All

Quick update. I've been working on some of the bugs I've found with this camera and will be documenting all of these to Dahua and @EMPIRETECANDY this week. Specifically 5 I'm tracking right now, all with relatively easy fixes that I've identified.
Also I will be working on another S3 review soon and therefore will hold on overall conclusion post until I've completed the series. I'm also doing some testing with these cams for LPR for you and will report back on my findings on the cam performance in those situations. Stay tuned

 

[lukeeper]

Thank you so much for usefull videos. Few questions:
1: Did you set gain as gain priority mode or manual mode? If manual - shutter should be 0-40?

1.5: In a DAY profile you are using mode - auto for exposure?

2. Gamma 42, contrast 52 and sharpness 40-44 are applied to both day and night?

3. Camera is working on customized scene and you are setting up obly day and night profile?

 

[Wildcat_1]

Thanks for the kind comment @lukeeper
Great questions. Let me expand with some additional answers for you

Q1- ALWAYS Manual Mode and dial in the range for your situation. Personally I don't like to go lower than 1/250 (absolute lowest) during the day. At night I'll go between 1/120 to 1/250 depending on install location and target capture needs
Q1.5 - NOPE, again ALWAYS use Manual Mode
Q2 - This depends on install location and target capture for this series of cams. ALSO depends on which MP can you purchase. for the 4K version (featured in this review), I would 100% keep Sharpness around 42 (split the difference) for day + night and keep Contrast around 52 (maybe 54) for both as well. Gamma is good at 42 during day BUT I've dropped this as low as 32 (yep, that low and even dipped to 28 depending on lighting at location vs vehicle caps) when using this series of cams for other applications like LPR (using the Z4 version of this 4K cam, more on this one later )
Q3 - I use customized and set day and night schedules for month over month / year. This is a HUGE benefit of the v3.0 FW and 5.0 GUI

HTH

Thank you so much for usefull videos. Few questions:
1: Did you set gain as gain priority mode or manual mode? If manual - shutter should be 0-40?

1.5: In a DAY profile you are using mode - auto for exposure?

2. Gamma 42, contrast 52 and sharpness 40-44 are applied to both day and night?

3. Camera is working on customized scene and you are setting up obly day and night profile?

 

[Wildcat_1]

Wanted to provide some updates. Had some issues getting onto the site this week but wanted to post my Tips/Recommendations here (will also post in the initial Post number) here. I am also getting ready to to edit the test for the next S3 camera but have been focused on getting the bugs documented (will post here too) that I've found so far so I can get those (and my proposed fixes) to Dahua ASAP. Lastly I am also going to be posting a separate video (using the S3 cams) for a challenge a couple of people DM'd me about, so watch out for that too.

 

[Wildcat_1]

Tips / Recommendations When Using The S3 Cameras (Listed in My Video too):

• Reduce Gamma to 42.
• Reduce Sharpness to 40 - 44.
• Reduce Exposure Compensation to 20.
• For Night Settings Consider Dropping Gain to Range of 0 - 40 (from 50).
• Increase Contrast to 52.
• At Night, Don't Drop Noise Reduction (2D or 3D) below 36 on These Cameras.
• Turn Off Auto Iris.
• Even Though AcuPick Needs Work, Turn It On To Get IVS + Video Metadata Captures tagged.
• If You Disable AcuPick You Only Get 1 or The Other E.g. IVS, Video Metadata, Face Detect.

Be sure to stay tuned for the challenge Tips & Recommendations where I specifically dive into an area using settings that may surprise a lot of you.

 

[cctvtech]

Hi Wildcat nice videos. I have used the 8MP out in the field and its great at getting faces and plates while zoomed in on a road (during the day). It gets them twice as far as the 4MP I also have on another site (eg. 600+ vehicles before 10am, 2000 human detections by midday). When I left it full screen it was catching cars passing on the main road and within 1 hour I had 60 pages of vehicles (disabled that quickly as I only needed the side road). It is very accurate though, better than expected. Eg. 150 feet away it will show me a white car in search result. I find the 8MP is not good for night though so I install it primarily for Daytime along with various model 4MP cameras for night.

So definitely the 8MP is best for daytime plates (non searchable) and faces with Metadata (when searching faces it shows results from face detection enabled channels plus Metadata enabled channels with face detect enabled). In fact on the XI series you can enable Metadata on a non AI camera and by default Face Detection is enabled in that "Metadata by Recorder" setting (can be disabled for low res cameras). XI lets you do 8 channels of AI By recorder (eg. 2x face recognition and 6 metadata, or 8 face recognition, etc. any combination). Im doing Face Recognition at various stores now using the 5442 ZE on the EI series NVR. In a couple cases I am using NON AI Cameras and Face Detection by Recorder. It all works much better than it did a few years ago. I tested Face Recognition alerts with custom databases and it worked very accurate at 80% setting even with low res faces in the databases.

XI series still has no Acupick in the web service though, waiting patiently as nobody will use a 2U rackmount on their desk to setup and use Acupick (all rack mounted or wall mounted locked away in IT rooms). It has to work over the Network.

That said, since you have Dahua's ear can you please ask them to make a LITE version of the camera webservice as that new animated one crashes the phone which is the only way to setup the cameras in many cases when dealing with big buildings and locked down IT departments. Even my faster PC has issues with it when setting up at home. Maybe they can use a toggle switch to LEGACY mode. As techs we dont need fancy animations. Thanks. Appreciated if possible.

 

[steve1225]

If You Disable AcuPick You Only Get 1 or The Other E.g. IVS, Video Metadata, Face Detect.

If You disable AcuPick and go into straight Video Metadata on 5442-S3, You have triple: Video Metadata, Face Detect (with face snapshots) and car plates snapshots (but not plate numbers - this can be only done on 7xx2-X)... Very functional for street / outside plot cams...

If You want full Video Metadata (with Face & Plate snapshots, without AcuPick) & IVS on public street cams and use Dahua 5xxx-EI NVR, You can disable AcuPick on NVR and use IVS done on recorder (up to 4 channels) for those street cams. In this configuration the rest cams (above 4) can have AcuPick (IVS + mini VMD) enabled and mini VMD with human attributes will work.

Disabling AcuPick on NVR simply disable base AcuPick 1.0 functionality (searching people/cars by IMAGE) not Video Metadata & human/cars attributes part (searching people/cars by attributes)..

There is third variant where You run full Video Metadata (with face & plate snapshots) on 5442-S3 cam, disable AcuPick on NVR and run Face Recognition (up to 8 cams) on 5xxx-EI NVR...

 

[steve1225]

So definitely the 8MP is best for daytime plates (non searchable) and faces with Metadata (when searching faces it shows results from face detection

You can try 7442-X..
It have full ANPR in Video MetaData mode (plates are OCR & searchable)..
Also have different video processing - I found that on 7442 I can setup low gain (like 30 max 40 for low noise) with short shutter for night mode & then do SSA/WDR which on 7442 will do second level gain, much better that 5442. 7442 works much better for motion at night conditions.

7442 (but only 4mpx version) have also time division exposure technology - which joins static/slow moving and fast moving objects from different shutter exposures.. something like WDR/HDR for motion...

XI series still has no Acupick in the web service though, waiting patiently as nobody will use a 2U rackmount on their desk to setup and use Acupick (all rack mounted or wall mounted locked away in IT rooms). It has to work over the Network.

Also wait for AcuPick 2.0 firmware for 6xx-XI...
Dahua promised that 6xx-XI will have option to do AcuPick on recorder & Face Detection/Recognition for AcuPick cams..
Release date for 6xx-XI promised by Dahua in AcuPick 2.0 presentation was July 2024.
But this didn't materialized.

ps. you can use HDMI over CAT5 extenders with KVM functions (transfer USB mouse back to transceiver)...

 

[cpd5215]

If You disable AcuPick and go into straight Video Metadata on 5442-S3, You have triple: Video Metadata, Face Detect (with face snapshots) and car plates snapshots (but not plate numbers - this can be only done on 7xx2-X)... Very functional for street / outside plot cams...

If You want full Video Metadata (with Face & Plate snapshots, without AcuPick) & IVS on public street cams and use Dahua 5xxx-EI NVR, You can disable AcuPick on NVR and use IVS done on recorder (up to 4 channels) for those street cams. In this configuration the rest cams (above 4) can have AcuPick (IVS + mini VMD) enabled and mini VMD with human attributes will work.

Disabling AcuPick on NVR simply disable base AcuPick 1.0 functionality (searching people/cars by IMAGE) not Video Metadata & human/cars attributes part (searching people/cars by attributes)..

There is third variant where You run full Video Metadata (with face & plate snapshots) on 5442-S3 cam, disable AcuPick on NVR and run Face Recognition (up to 8 cams) on 5xxx-EI NVR...

You seem to have a very good working knowledge of Metadata and AcuPick so I figured I would ask you. I bought a NVR8CH-8P-2AI which I believe is 5 series??? It has AcuPick. I also have several 5442-S3 cams and Z4-S3's. With that said, with the settings I have enabled (see screenshots) I have 3 questions.

1.) When I search the NVR for "Face detection", it produces no results. I would like the 5442 by my front door to detect faces and then those captures by easily searched by the NVR. I dont necessarily need this ability on the rest of my cams. I use IVS with people, vehicle, and animal detection on ALL cams (setup on cam itself) as of now which works really well. It appears in the NVR GUI that Face Detection is turned on.

2.) I have one interior cam, IP2M-841 connected to NVR via WIFI that I have dumb motion detection enabled only when we go to bed. Is there anyway for the NVR to impose SMD on this cam since the cam just does all motion and I get a lot of false triggers in the middle of the night. I cant seem to figure this out and maybe it has something to do with AcuPick enabled on NVR ( I really like the power of AcuPick).

3.) lastly, I think I read in the NVR manual but cant find it now, but I thought there was the abilty to report "false" images or somehting like that to the NVR database when AcuPick or metadata was captured and it was the "wrong" item in an effort to improve the false triggers. This is not a big deal but thought I would give it a whirl and see if it helped.

Thanks again. I enjoy reading you and others posts with extensive knowledge of these NVRs and cams.

 

[steve1225]

You seem to have a very good working knowledge of Metadata and AcuPick so I figured I would ask you. I bought a NVR8CH-8P-2AI which I believe is 5 series??? It has AcuPick. I also have several 5442-S3 cams and Z4-S3's. With that said, with the settings I have enabled (see screenshots) I have 3 questions.

it's 5208-8P-EI...

1.) When I search the NVR for "Face detection", it produces no results. I would like the 5442 by my front door to detect faces and then those captures by easily searched by the NVR. I dont necessarily need this ability on the rest of my cams. I use IVS with people, vehicle, and animal detection on ALL cams (setup on cam itself) as of now which works really well. It appears in the NVR GUI that Face Detection is turned on.

NVR allows to setup all AI options done on cam (IVS, VideoMetaData, Face Detection, Face Recognition) without checking which AI function is selected on Smart Plan on that camera. But this don't mean that AI function will work - Smart Plan selection done on camera is valid.

If You want Face Detection on 5xxx-S3 camera, then the best solution is disable AcuPick on that camera and select & configure Video Metadata on that camera. You will get human/cars detection with attributes (age/sex/colors) plus face detection & car plates snapshots.. You will be able to browse faces or humans or cars in NVR interface (AI search -> Face detection & AI search -> Human detection & AI search -> motor vehicle detection)..

You will lose IVS functionality on that camera.

If You need IVS for that cam, You must disable AcuPick on NVR side (AI settings -> Smart Plan -> Recorder tab -> AI Mode).
In that case You will lose ability to search people / cars by IMAGE (plus AI search -> AcuPick menu) but all functionality from Video Meta Data (human / face / motor search by attributes described above) will be working.

After reboot You will get ability to configure IVS by 'AI on recorder' for that channel (configuration only by NVR interface).

2.) I have one interior cam, IP2M-841 connected to NVR via WIFI that I have dumb motion detection enabled only when we go to bed. Is there anyway for the NVR to impose SMD on this cam since the cam just does all motion and I get a lot of false triggers in the middle of the night. I cant seem to figure this out and maybe it has something to do with AcuPick enabled on NVR ( I really like the power of AcuPick).

the same - You must disable AcuPick on recorder.
then You will have ability to run SMD (8 channels) / IVS (4) / Face Detection (2) &recognition (8 channels if face detection or full video meta data is enabled on cam) by 'AI on recorder'.

AcuPick on recorder and all other AI modes on recorder don't work together on 5xxx-EI.
For that You need higher line of NVR (6xx-XI)..

3.) lastly, I think I read in the NVR manual but cant find it now, but I thought there was the abilty to report "false" images or somehting like that to the NVR database when AcuPick or metadata was captured and it was the "wrong" item in an effort to improve the false triggers. This is not a big deal but thought I would give it a whirl and see if it helped.

this is 5xxx-S3 function on latest firmware.
You will find 'self lerning' option in global configuration in IVS settings (both on NVR or cam).
After enabling there will be option for editing experience database, where You can upload 'false positives' (cutted images of things which creates false positives - like tress etc)..

You can also restrict area (in video frame) in Video Meta Data configuration for detection of people / cars / non motor vehicles to exclude problematic frame fragment (like trees or public street with high cars traffic).

 

[cpd5215]

it's 5208-8P-EI...




NVR allows to setup all AI options done on cam (IVS, VideoMetaData, Face Detection, Face Recognition) without checking which AI function is selected on Smart Plan on that camera. But this don't mean that AI function will work - Smart Plan selection done on camera is valid.

If You want Face Detection on 5xxx-S3 camera, then the best solution is disable AcuPick on that camera and select & configure Video Metadata on that camera. You will get human/cars detection with attributes (age/sex/colors) plus face detection & car plates snapshots.. You will be able to browse faces or humans or cars in NVR interface (AI search -> Face detection & AI search -> Human detection & AI search -> motor vehicle detection)..

You will lose IVS functionality on that camera.

If You need IVS for that cam, You must disable AcuPick on NVR side (AI settings -> Smart Plan -> Recorder tab -> AI Mode).
In that case You will lose ability to search people / cars by IMAGE (plus AI search -> AcuPick menu) but all functionality from Video Meta Data (human / face / motor search by attributes described above) will be working.

After reboot You will get ability to configure IVS by 'AI on recorder' for that channel (configuration only by NVR interface).



the same - You must disable AcuPick on recorder.
then You will have ability to run SMD (8 channels) / IVS (4) / Face Detection (2) &recognition (8 channels if face detection or full video meta data is enabled on cam) by 'AI on recorder'.

AcuPick on recorder and all other AI modes on recorder don't work together on 5xxx-EI.
For that You need higher line of NVR (6xx-XI)..



this is 5xxx-S3 function on latest firmware.
You will find 'self lerning' option in global configuration in IVS settings (both on NVR or cam).
After enabling there will be option for editing experience database, where You can upload 'false positives' (cutted images of things which creates false positives - like tress etc)..

You can also restrict area (in video frame) in Video Meta Data configuration for detection of people / cars / non motor vehicles to exclude problematic frame fragment (like trees or public street with high cars traffic).

Well, shit I figured you would know one or two of these but not all three. LOL. Thanks a lot for the detailed explanation of all of these items. That will help me a lot.

 

[bigredfish]

I’ll second the use of Metadata on the S3s. I have it enabled on my close up street view and am able to get face detection, human, vehicle, and non vehicle results on my older 5216-16p-4Ks2E NVR. All of that on a single camera even using an older NVR is quite powerful

 

[steve1225]

Well, shit I figured you would know one or two of these but not all three. LOL. Thanks a lot for the detailed explanation of all of these items. That will help me a lot.

heh..
I do all those things on all installed by me systems..
full cars / plates / human / faces database with attributes for dedicated street / gate cams (AcuPick disabled & full version of Video MetaData done on camera) plus cars / human database & IVS for inside plot / house cams (AcuPick enabled with mini IVS & mini Video MetaData)..

one / two 5442-Z4 (or 7442-Z4) zoomed cams dedicated for street / gate can generate thousands of cars / humans records per day..

 

[cpd5215]

heh..
I do all those things on all installed by me systems..
full cars / plates / human / faces database with attributes for dedicated street / gate cams (AcuPick disabled & full version of Video MetaData done on camera) plus cars / human database & IVS for inside plot / house cams (AcuPick enabled with mini IVS & mini Video MetaData)..

one / two 5442-Z4 (or 7442-Z4) zoomed cams dedicated for street / gate can generate thousands of cars / humans records per day..

Just so I understand this correctly, you can enable or disable AcuPick per individual camera and not just on for ALL cams connected to NVR or all off? Also what’s “mini IVS”. Lol. I’m sure I’ll recognize what you mean once you explain it a little bit.

I have to say, these cameras with the firmware upgrades are really quite impressive. I’m the typical story guy that come from big box brands just because of their ease of use, but they had shitty quality, especially at night as you guys constantly remind us of. These cameras and NVR are quite powerful devices.

Thank you.

 

[steve1225]

Just so I understand this correctly, you can enable or disable AcuPick per individual camera and not just on for ALL cams connected to NVR or all off? Also what’s “mini IVS”. Lol. I’m sure I’ll recognize what you mean once you explain it a little bit. Thank you.

Normally (historically) You can do ONE AI function (one AI algorithm) on camera.
You must choose one You are interested by selecting proper SmartPlan.

AcuPick 2.0 is first time in Dahua mainstream cams, that You have 2 mainstream AI models / functions running together on one camera.
In case of AcuPick 2.0 this is IVS (perimeter protection) & Video MetaData (people/car capture / descriptions by attributes).
But due SOC (CPU) power limitations, both IVS & Video MetaData are running in limited versions (comparing to standalone mode without AcuPick, where You select only IVS or only VMD).

IVS is limited only to tripwire/intrusion and You don't have long distance mode.
Video Metadata is limited only to full human/car snapshots (so no plate / face capture) and some attributes on cars are missing.

You can enable/disable AcuPick per camera. And also have cams without AcuPick functionality.
On some (mostly internal) cams is better to have mini IVS & mini VMD working together (so you enable AcuPick).
On some (mostly street/gate/outside plot) is better to have full version of VMD without any IVS (or with IVS done on recorder).

Independent of enabling/disabling AcuPick on cams, you can enable/disable AcuPick on recorder.
Recorder AcuPick (aka AcuPick 1.0) is doing only 3D modeling on humans / cars - so You can search human / cars by IMAGE.
Clicking on AcuPick icon in live/recorded NVR view and then on any human or car on image NVR will search for all "SIMILAR" looking humans / cars from on all footage from all AcuPick enabled cams. So it's very easy to find the same person from other cams or from past footage.

Recorder AcuPick mode require cams with AcuPick enabled. And blocks all other AI functions / models run on recorder (AI by recorder) on 5xxx-EI..

By disabling AcuPick on cam or recorder You loose only 3d-modeling part (searching by IMAGE).
All cam IVS or all cam VMD functionality is working (they don't use any AI on recorder).

By disabling AcuPick on recorder You gaining ability to run AI models / functions on recorder for dumb cams / intercoms.
Or add second AI function done on recorder to existing one done on cam - like adding IVS or Face Recognition done on recorder to VMD done on camera.

 

[cpd5215]

Normally (historically) You can do ONE AI function (one AI algorithm) on camera.
You must choose one You are interested by selecting proper SmartPlan.

AcuPick 2.0 is first time in Dahua mainstream cams, that You have 2 mainstream AI models / functions running together on one camera.
In case of AcuPick 2.0 this is IVS (perimeter protection) & Video MetaData (people/car capture / descriptions by attributes).
But due SOC (CPU) power limitations, both IVS & Video MetaData are running in limited versions (comparing to standalone mode without AcuPick, where You select only IVS or only VMD).

IVS is limited only to tripwire/intrusion and You don't have long distance mode.
Video Metadata is limited only to full human/car snapshots (so no plate / face capture) and some attributes on cars are missing.

You can enable/disable AcuPick per camera. And also have cams without AcuPick functionality.
On some (mostly internal) cams is better to have mini IVS & mini VMD working together (so you enable AcuPick).
On some (mostly street/gate/outside plot) is better to have full version of VMD without any IVS (or with IVS done on recorder).

Independent of enabling/disabling AcuPick on cams, you can enable/disable AcuPick on recorder.
Recorder AcuPick (aka AcuPick 1.0) is doing only 3D modeling on humans / cars - so You can search human / cars by IMAGE.
Clicking on AcuPick icon in live/recorded NVR view and then on any human or car on image NVR will search for all "SIMILAR" looking humans / cars from on all footage from all AcuPick enabled cams. So it's very easy to find the same person from other cams or from past footage.

Recorder AcuPick mode require cams with AcuPick enabled. And blocks all other AI functions / models run on recorder (AI by recorder) on 5xxx-EI..

By disabling AcuPick on cam or recorder You loose only 3d-modeling part (searching by IMAGE).
All cam IVS or all cam VMD functionality is working (they don't use any AI on recorder).

By disabling AcuPick on recorder You gaining ability to run AI models / functions on recorder for dumb cams / intercoms.
Or add second AI function done on recorder to existing one done on cam - like adding IVS or Face Recognition done on recorder to VMD done on camera.

Again thanks a lot for the detailed explanation. I think this is starting to make sense to me. I’ll screw around with it for a little while and see what I like best. I appreciate the help and suggestions.

 

[Wildcat_1]

Yes the S3 cams are very capable and offer functionality and flexibility that allow the use of Acupick vs vanilla IVs on a camera by camera basis. The addition of Video Metadata as I mentioned up front, is a key win on these and other cams that feature it. If you have AI enabled NVRs like 5216-XI then you can hand that any stream and let the NVR to the processing with processing by NVR which works very well as I’ve shown before (and will share updates here too) BUT having the ability for on-camera processing is incredibly beneficial and certainly reduces ROI when deploying these cams.

That’s not to say the S3 doesn’t have issues as I’ve mentioned, I’m tracking 5 bugs already. It’s also not to say that AcuPick 2.0 also doesn’t have issues (have additional bugs and recommended code level changes for that too) BUt the S3s are a BIG step in the right direction.

The 7xxx series is also very capable (reviewed, used and deployed many of those too) BUT they also have a place.

The beauty of the 5x S3’s is they are affordable, compatible with PFA square mounting boxes for easy reuse in current deployments and as I co time to mention, incredibly capable.

As I’ve shown in my review, continue to advocate for, and help people with, is to dial in these cams. The S3s need more dialing in to get the right balance but when you do, the results speak for themselves. Stay tuned.

 

[Wildcat_1]

Normally (historically) You can do ONE AI function (one AI algorithm) on camera.
You must choose one You are interested by selecting proper SmartPlan.

AcuPick 2.0 is first time in Dahua mainstream cams, that You have 2 mainstream AI models / functions running together on one camera.
In case of AcuPick 2.0 this is IVS (perimeter protection) & Video MetaData (people/car capture / descriptions by attributes).
But due SOC (CPU) power limitations, both IVS & Video MetaData are running in limited versions (comparing to standalone mode without AcuPick, where You select only IVS or only VMD).

IVS is limited only to tripwire/intrusion and You don't have long distance mode.
Video Metadata is limited only to full human/car snapshots (so no plate / face capture) and some attributes on cars are missing.

You can enable/disable AcuPick per camera. And also have cams without AcuPick functionality.
On some (mostly internal) cams is better to have mini IVS & mini VMD working together (so you enable AcuPick).
On some (mostly street/gate/outside plot) is better to have full version of VMD without any IVS (or with IVS done on recorder).

The Video Metadata streams actually can contain plate information. It’s down to implementation of video metadata that makes the difference. Therefore depending on cam, FW implementation etc there's not always a need for using NVR processing (for NVR's that feature it). There are Dahua cams with Video Metadata tagged streams that do carry plate information vs others where this is not featured.
Even on those that don’t enable plate as part of VMD you still see cams correctly tag those caps as ANPR BUT the functionality of plate is stripped. This is done a lot of time to not cannibalise the higher end models such as 7xxx or even more so the true LPR traffic cams that I’ve reviewed before.
A good example are PTZ’s. Early iterations of the 49xxx cams that I reviewed had plate information as part of Video Metadata tagged streams BUT later FW removed that ability. This is a reason I advocate for and still use, older FW on those deployments.

Also the benefit of the recent additions to the S3 series is that through IVS AI (AcuPick mode or Standalone) you gain, on camera learning that you can influence which is another huge benefit not seen before.

Yes, I agree it is key to setup databases for Face, Vehicle etc as part of a successful deployment AND a complete surveillance solution. I’ve shown this in the Face database overviews in some of the cam reviews I’ve done.

The key part here is that with these newer cams we’re getting to an exciting inflection point of needed features, implemented with improved optics and SOC pairings at affordable prices with little compromise. This means that a number of the barriers for entry for most people, business, commercial deployments are being removed, that’s a good thing

VMD is something I’ve advocated for and shown the benefits of for a long while and it’s great to see this feature across more mid level cams now.

 

[Wildcat_1]

Adding in an EI NVR into this mix is a big benefit. If you want to go further then using something like an XI (which I use in addition to EI) takes it to another level. All depends on use case, budget and your choice of ecosystem for NVR, with its middleware based, software NVRS (BI etc) or standalone.

Another note (and plus point) about the S3's for those that weren't aware is that when you choose the Video Metadata function (AcuPick or Standalone) they do show up with ANPR functionality. Now, with that said, this is not implemented well yet and a bug I've been speaking with Dahua about BUT critically it IS supported by these cams. The Dahua NVR's will pickup on this attribute as well. However current (including the latest September) FW has not enabled the functionality correctly. I'm hoping Dahua will stand behind this and correctly implement soon. As I said, I've provided a ton of information and recommendations on this, it's now on them.

 

[steve1225]

The Video Metadata streams actually can contain plate information. It’s down to implementation of video metadata that makes the difference. There are Dahua cams with Video Metadata tagged streams that do carry plate information vs others where this is not featured.

Yes, I have a few 7442-Z4-X in different locations...
Works very well for ANPR duties in VMD mode (in color at day, in IR mode at night with very well visible cars)..

Even on those that don’t enable plate as part of VMD you still see cams correctly tag those caps as ANPR BUT the functionality of plate is stripped. This is done a lot of time to not cannibalisthe higher end models such as 7xxx or even more so the true LPR traffic cams that I’ve reviewed before.

5xx2-S3 is doing that.. sending plate snapshot but there is no plate OCR..
yes, they don't do ANPR on 5xxx-S3 to not cannibalize 7442...
But also due more limited SOC power in 5xx2-S3...

Also 7442-X don't have normal AI functionalities like AI-SSA or AI-Codec.
This is done to spare one of the NPUs (AI processors) in HiSilicon SOC in 7442-X.

HiSilicon SOCs usually have two almost identical NPUs processor to run 2 different AI models in parallel.
In 5xx2-S3 one is used to run AI-SSA / AI-Codec models and second one to run selected in Smart Plan AI algorithm.

in 7442-X two NPUs are free to run selected in Smart Plan functions - it can run 2 different smaller algorithms (like Face Detect + Face Recognition) or one bigger like ANPR or Video Meta Data (probably OCR part is run on second NPU core)..

The key part here is that with these newer cams we’re getting to an exciting inflection point of needed features, implemented with improved optics and SOC pairings at affordable prices with little compromise. This means that a number of the barriers for entry for most people, business, commercial deployments are being removed, that’s a good thing

1. there are compromises - there are lots of bugs in firmware & implementation. Of course every generation of firmware or every generation of cams fixes some, but also adding a new ones...

2. Actual generation of cams (5xx2-S3, 7xx2-X) and NVR (5xxx-EI, 6xx-XI) added possibility to run more that one AI algorithm - which gave us much more functionality outside simple IVS. But there are still huge limitations which functions / algorithm work together with which one in what configuration or what functionality we are losing when enabling another one.

I hope that next gen of cams & NVR will solve that.
Running IVS + AcuPick 1.0 (3d modeling) + Video Metadata with attributes + Face Detect/Recognition + ANPR on one cam should be simply functionality available without any limitations on PRO camera / NVR.