Is there much difference between the 5232EI and NVR8CH-AI?
If I read correctly does or only support FD from 2 cameras?
Are faces cropped or zoomed and cropped? I have a couple IPC-T54IR-ZE but will probably get 2 more.
Help me understand the difference between Face Detection and Face Recognition
- Thread starter Bill Gates
- Start date
If I read correctly does or only support FD from 2 cameras?
Are faces cropped or zoomed and cropped? I have a couple IPC-T54IR-ZE but will probably get 2 more.
All Andy NVRs with ending -AI (-2AI, -4AI) are 5xxx-EI..
the same NVR, only difference max camera limit, HDDs and POE/non POE versions...
If You want buy 8CH non POE version, look at 16CH or even 32CH.. 20$ difference between each, but You will not be limited by camera count..
All 5xxx-EI have limit for 2 ch face detection/recognition (if all is done on NVR) or 8 ch face recognition (if face detection is done on camera).
if you enable face detection of full video meta data, you loose IVS on camera (but it can be done on NVR).
Generally: 5xxx-EI is little limiting if you really want to play with face detection/recognition.
Much higher AI power / limits are on 6xx-XI (-XI models on Andy shop) but they cost much much more..
the same NVR, only difference max camera limit, HDDs and POE/non POE versions...
If You want buy 8CH non POE version, look at 16CH or even 32CH.. 20$ difference between each, but You will not be limited by camera count..
All 5xxx-EI have limit for 2 ch face detection/recognition (if all is done on NVR) or 8 ch face recognition (if face detection is done on camera).
if you enable face detection of full video meta data, you loose IVS on camera (but it can be done on NVR).
Generally: 5xxx-EI is little limiting if you really want to play with face detection/recognition.
Much higher AI power / limits are on 6xx-XI (-XI models on Andy shop) but they cost much much more..
If you ask for 6xx-XI, then:
empiretech01.com
empiretech01.com
EmpireTech NVR32CH-8XI 32 Channels 2U 8HDDs Network Video Recorder
Support multiple decoding formats Max. decoding capability: 32 × 1080p@30 fps Supports AcuPick with up to 32 channels AI by Recorder AI by Camera Supports N+M cluster Raid 0/1/5/6/10, ISCSI Allows for HDD hot swapping Supports EPTZ One-click Arming & Disarming Security baseline 2.3
empiretech01.com
EmpireTech NVR64CH-8XI 64 Channels 2U 8HDDs Network Video Recorder
Support multiple decoding formats | Max. decoding capability: 32 × 1080p@30 fps | Supports AcuPick with up to 64 channels | AI by Recorder | AI by Camera | Supports N+M cluster | Raid 0/1/5/6/10, ISCSI | Allows for HDD hot swapping | Supports EPTZ | One-click Arming & Disarming | Security...
empiretech01.com
Got it. So face detection would give you a snapshot of the face. I suppose that's okay, saves you some time instead of you making the snapshot yourself.
So if I have 2 of the new 5442 ZE and looking to add 1 or 2 more along with 1 PTZ5A4M-25X, would the 5xxx XI series be sufficient for Facial Recognition like your thumbnails along with still using IVS/Tripwire for alerts? Along with 2-3 of the IP2M-841.
Total of 4-5 Smart Cameras and 2-3 of the Amcrest Indoor.
I don't think I'd ever use the heat map feature and am unsure if smart object detection would be needed since human detection works great with IVS right now.
Or would I still need the 6xxx series? if so is there a cheaper 8ch or 16ch model? Are there any video walk-throughs of the 5xxx XI or 6xxx series? Would like to see more of the features before dropping 1k
As an Amazon Associate IPCamTalk earns from qualifying purchases.
bigredfish
Known around here
On each camera you will have to choose to either run Face Detection OR IVS. You can't run both on the camera simultaneously.
The 5442 does not have Face Recognition
So if you want to run Face Recognition you would need to run that on the NVR (2 channels max on the 5xxx series, 8 channels max on the 6xxx series 16ch unit) and IVS on the cameras.
Number of AI channels on the NVR varies depending on how many total channels it has. Its in the descriptions
empiretech01.com
*If you don't need Recognition, but like the idea of AI search with thumbnails of faces and can live with Detection, then the 5xxx series is all you need.
You can use 1 AI function per camera again. Either FD, IVS, or VMD.
The benefit of VideoMetaData function is it gets you 4 AI models in one AI function as I showed which is pretty cool.
**Not every scene or every camera is going to be ideal for face detection/recognition. Its not as sophisticated as you see on TV. It works best closer to the camera FOV and with good lighting etc. Settings such as exposure are still important to be dialed in to get best results.
The 5442 does not have Face Recognition
So if you want to run Face Recognition you would need to run that on the NVR (2 channels max on the 5xxx series, 8 channels max on the 6xxx series 16ch unit) and IVS on the cameras.
Number of AI channels on the NVR varies depending on how many total channels it has. Its in the descriptions
Network Video Recorder | EmpireTech
EmpireTech has a range of recorders from the Lite Series (4CH/unit) for smaller uses, the Pro Series (optimal for mid-to-large projects) to the Ultra Series (256CH/unit) for expansive projects, plus cost-effective options.
empiretech01.com
*If you don't need Recognition, but like the idea of AI search with thumbnails of faces and can live with Detection, then the 5xxx series is all you need.
You can use 1 AI function per camera again. Either FD, IVS, or VMD.
The benefit of VideoMetaData function is it gets you 4 AI models in one AI function as I showed which is pretty cool.
**Not every scene or every camera is going to be ideal for face detection/recognition. Its not as sophisticated as you see on TV. It works best closer to the camera FOV and with good lighting etc. Settings such as exposure are still important to be dialed in to get best results.
As a @bigredfish wrote, You can do only one AI function on camera (by selecting SmartPlan).
So You can't mix IVS and Face Detection...
there is one exception - 5442-s3 with latest firmware when You enable AcuPick mode, it enables IVS and mini Video Meta Data.
in mini VMD there are no face or car plates snapshots, but human & vehicle snapshots are done (and searchable by NVR).
But this can't be used for face recognition on NVR.
with 5xxx-EI you are limited to 2 cameras with face detection/recognition, when You do IVS on them
with 6xx-XI you can do that on 8 cameras...
So You can't mix IVS and Face Detection...
there is one exception - 5442-s3 with latest firmware when You enable AcuPick mode, it enables IVS and mini Video Meta Data.
in mini VMD there are no face or car plates snapshots, but human & vehicle snapshots are done (and searchable by NVR).
But this can't be used for face recognition on NVR.
with 5xxx-EI you are limited to 2 cameras with face detection/recognition, when You do IVS on them
with 6xx-XI you can do that on 8 cameras...
Hi Steve.
How exactly is this setup? I’ve ordered a B54IR-Z4E-S3 from Andy during the recent sale to use for license plate capture. I’m interested in getting the same setup here. Is that just going into the camera and turning on video metadata with Acupick off? Or can you have Acupick on with IVS and video metadata. Then you search on NVR by AI Search to view?
Sent from my iPhone using Tapatalk
bigredfish
Known around here
May depend on your NVR model, but on my older 5xxx series yes, I simply choose Video MataData without AcuPick and set it up like any AI rule.
Then choose the attribute in the NVR AI search (Human, Face, Vehicle, NonMotorVehicle) and it will give you results for each as we showed
Then choose the attribute in the NVR AI search (Human, Face, Vehicle, NonMotorVehicle) and it will give you results for each as we showed
As I wrote many times - you must select beetwen: Acupick turn on with only mini IVS and only mini VMD (no face/plate snapshots, many limitations) or AcuPick turn off for full VMD with all features (as presented on my post) and no IVS (or IVS done on NVR)...
Last edited:
bigredfish
Known around here
Appreciate the help here. I was getting my self confused on the relationships/settings between NVR and camera. I’m going to set this up. Looks like it will be very helpful for my new camera I’ll be adding in.
Sent from my iPhone using Tapatalk
Thanks for the info and settings. I’ve got the a 5432 NVR so UI is familiar. I’m going to play with this. Was reading over this post and some others and got a bit confused on vmd vs mini vmd as well as setting up IVS. This helps me understand it more.
Sent from my iPhone using Tapatalk
[mention]steve1225 [/mention] Read back over this post after running into a problem. Hoping you might have some insight.
I’ve got 2 cameras (T58IR-ZE-S3) setup now running full VMD on each camera, with IVS running on the NVR. All is working great.
I’ve got a PTZ425DB-AT running IVS on camera. Also works great. Here’s my problem. When I got to turn on Face Detection for this camera on the recorder, I get an error. It states “Limited resources. Please disable IVS of recorder first and then enable the function”.
I’ve got Andy’s NVR32CH-4AI where I see in specs it supports 2 channels of face detection at recorder and 4 channels of IVS detection at recorder. I’m using 2 of the 4 IVS channels on the recorder and nothing else. I should have 2 more IVS channels to use and 2 Face Detection channels on the recorder per the spec.
This is not making sense to me. Any ideas? Appreciate any assistance.
Sent from my iPhone using Tapatalk
bigredfish
Known around here
Are you saying that rule applies to the NVR, AI by recorder for IVS and Face Detection? I get it on the camera level but specs for NVR make it seem that there is a channel count for each type IVS and FD.
I played with it and turned off IVS by recorder on one camera running VMD by camera. I was then able to turn on FD by recorder for my PTZ with IVS on camera with no errors. Both work well. Now if I go back to the 2nd camera running VMD and try IVS by recorder for that it won’t allow.
Sent from my iPhone using Tapatalk
bigredfish
Known around here
Thanks. I might have to ping Andy as his spec description is definitely misleading. What’s odd is I’m currently doing both with the last change I mentioned. I just can’t get 2 cams IVS by recorder with 1 FD by recorder.
Sent from my iPhone using Tapatalk