Hanwha Q-Series AI Cameras Tested
Hanwha Vision released its Q-Series AI cameras, which now add people and vehicle detection analytics on their mid-range cameras but how do they perform in real-world scenarios, and how do they compare to Hanwha's higher-end X or P-Series analytics?
We bought and tested the Hanwha QNV-C8011R, examining:
- How accurate are the Hanwha Q-Series Analytics vs. the P-Series analytics?
- How accurate are the Hanwha Q-Series Analytics vs. Axis Object Analytics?
- How accurate are people or vehicle detection in day or night scenes?
- How accurately do Hanwha Q-series analytics handle people and vehicle classification?
- How accurately does Hanwha Q-Series ignore common indoor and outdoor false alert sources (rain, light changes, foliage, headlights, animals, etc.)?
- How accurate are people detected in different postures (crawling, running)?
- How accurate are vehicle type classifications(bus, truck, car, bike)?
- How does WDR performance compare to the last generation non-AI Q-Series Camera?
- Does WiseNR decrease noise and improve image quality in a low-light scene?
- How does this compare to Axis camera analytics?
This is part of our benchmarking test series: Camera Analytics Rankings 2023 - 21 Manufacturers, 40 Analytics.
*******
*** ****** *-** ******* *** ~*/* the **** ** *** *********** **** P ****** *******. *** *********-*********** *** ~$*** ****, ***** ******* *****-******* *** ~$*** **** *** ************ ** ****** *********.
Physical ********
** *** ********* *****, ** ******* a ******** ******** ** *** *********-******
Executive *******
****** ***-****** ******** ** ******* ******/******* detection, ******** ****** ****** *** ******* sources ** ***** ****** *** ********** classifying ****** *** ********, ****** ** was ******** ** ***** **** ****** detection *** ******* **** ********* ******** to ******'* */*-****** ****** *********.
**** ***** *** ********* ******* ** our ******* ** *** *** ****** QNV-C8011R:
- ***% ** ****** **** ******** ** a ******** ** *****/** *******, ******* they **** **********, ********, ** *********. The *-****** ********* ******** ** ** ~170’ **** * ***° ***, ***** the *-****** ******* ***** *** **** with * ***° ***. *******, ****** these **********, *** ********* **** *** both ******* ** ~**% *** **** series. **** *** ****** * *** Q-Series ********* ******* ** * ******* resolution ****** **** *** ****** ** the ******** **** *** ****** ****** resolutions.
- ~**% ** ****** **** ******** ** night ** *** * ****** ******, which *** ******** ******** ** *** 90% ****** ********* **** ** *** P-Series. *** ****** **** ****** *** to ****** ***** ********** ** ******* and *** ***** ******* *********** ** night.
- ***% ** ****** ******* ** ****** and ******* ***** ****** **** *******, this ******** ******* *******, ******** ********, shifting ******* ********** ***** *****, ********** rain, *** ******* ******* **********.
- ***% ** ****** *** ******** **** classified ********* *** *** *** ******** people *** ********.
- ~**% ** ******* ***** **** ********** correctly ** *** *-****** *********, ******** to *** *-****** ********* *********** ~**%.
*** ****** ****** *-****** ****** (***-******) WDR ******** **** ******* ** * WDR ***** **** *** **** ********** non-AI *-****** (***-*****). ****** ********* *** noise ** * ***-***** ***** *** increased ******** ************* *** *** *** offer ******* ****** ** * ******* in ******.
Vs ****** *-****** *********
*** ******-*** ****** *-****** ******* ******** higher ********* ******** ******* ** *** tests, *** **** ********** ********** **** the *** *-****** **** **, **** as ********/******* *****, ****** **********, ***., but *** *-****** **** **** ****.
******** ** *** *-****** *********, *** Q-Series ** ****** ********* ******** ********* in * ******* ********, ********* ***% of ****** ** *** *****. *******, at *****, *** *-****** *******' ******* person ********* **** ***** ** ~**%, slightly **** **** *** *-****** ******* person ********* **** ** ~**%.
***** **** *** * *** *-****** cameras ******* ******* **** *********, *** P-Series ******* ********** ******* **** ********** ~75% ** *** ****, ***** *** Q-Series ****** ********** ******** ******* **** ~50% ** *** ****.
Axis ****** ********* **********
**** ****** ********* ************ ****** *-******, though ** ***** *** ****** *****. Axis ****** ********* ******** * ****** person ********* **** ** ***-***** *** more ******** ******* **** *************** ** our *******. *******, **** *** ****** with ***** ****** ** *******, ***** Hanwha *-****** ** *** ***.
"BestShot" **. "*************"
****** *-****** ******* ******* "********," ***** is ********* **** *** *-****** "*************," the *-****** ***** * **** ** the ****** **** *** **********, ***** the *-****** ***** * **** ** the ***** ***.
*** *-****** ******** ******** **** *********** in *** ***** ***** **** *** alerted **, ***** ***** ********* ******* easier, ***** *** *-****** ************* ***** you *** **** *****, ***** **** cause ***** ** ******* *** **** frame, *** **** *** **** ** obvious **** *** ******* **.
*****
***** ** ****** *-****** ********* ** a ****** *******, ***** ******** ********** into *** ********* ***, ******* ****** detection **, *** *********** *** ******* that *** ********. ** ***** *** setup ** *** ***** *****.
Scene ****** ******** ** ***** **********
*** ***** ******** ** * ***** resolution **** *** ******* ***, ***** means *** ******'* ********** **** *** determine *** ********* ********. ** *** testing, **** *** ***° ***, *** QNV-C8011R ******** ****** ** ~***'.
100% ** ****** ******** ** ******* *** ***** (**) *******
** *** *******, ***% ** ****** within *** ~***' ********* ******** ** the ***-****** **** ******** *** ******* on, **** ** ********* **** ** the ***** ***** ***** *** ****** is ******* ** * ******* **** most ** ***** **** ********** *** behind * ******.
Night ****** ********* **** ******* ** ~**%
** *****, *** ****** ********* **** dropped ** ~**%, ****** *** ** lower ***** ******, ***** ** *** provide * **** **** ** * person.
********* *********** *** *** ******** *** rate ** ****** *******.
100% ** ****** ******** *******, ********, **********
****** *-****** ********* ******** ****** ******* or ******** ** *** ***** ***% of *** ****.
****** ********* ********** **** **** ******** and ******* **.
100% ** ******* ******* ** ***** ****** *******
*** ****** *-****** ********* ******* *** outdoor ******* ** ***** ******, ********* one ** *** **** ****** ******* of ***** ******, ***** ****.
****, ***% ** ******* ** *** test **** ********* ******* / *** alerted.
******* ******* **** *** *** ***** false ******.
*******, ******* ********** ****** ******* *** scene **** ********* *******.
100% ** ****** ****** ******* ** ***** ****** *******
****** *-****** ********* ******* *** ****** sources ** ****** ***** ******, **** as ***** ***** *******.
************, ****** ******* **** **** *******.
100% ** ****** *** ******** ********* **********
*** ****** *-****** ********* ********* ********** all ****** ** ****** *** *** vehicles ** ********.
50% ** ******* **** ********** *********
*******, *** ****** *-****** ********* ********* classified *** **** ** ******** **% of *** ****.
*** **** ****** ****** **** **** in *** ***** ***** ********** ** trucks *** ****.
****** ******** ** **** * *** should ** ******** ** * ***, though ** ***** ** ********** ** a ***** **** ******* ******.
* *** ****** ** ******** ** a ***, ******* ********* ** *** shape ** *** *** (*****, *******, etc.), *** *** ***** ** *** vehicle ** ******** ** *** ******, it ***** ** ******** ** * truck ** **** *****. ** **** attempt ** ******* *** ** ***** regarding **** *** *** ******. ** are ***** ** ** ********** ** email ******* *** ****** ** *********/********* detection. ** **** ** ****** ****** to **** **** ********* *** *-** second ***** *******. ** **** ******* this **** *** **** *** ** lab **-***** *** ***** *******. ** plan ** ****** *** ** ****** a *** ***** *** **** **** this ******** **** ***** ******* **** our *&* **** & ** ***. I **** **** *** ***** ******* for **** ***** *****, *** ** will **** ** ****** ** * page ** *** ******* ** *** next *** ** ***. **** ***** address *** ** ******** ** *** report *** *** ***** ** **** feedback.
Less **** *% ** ******* ********** **** *********
****** *******, ********* ** *** ***** were ********** ** ********, ****** **** happened **** **** *% ** *** time.
****** *********:
********* *** ******** ** **** ******* data *** **-***** *** ** ***** to ***** ***** ***** **********. ** you *** ******* * ***** ****, such ** ** ******* **** *** camera ******* *** ******** (****** ***** in ****), **** ***** ** ****** and ** *** ******* ** ** the **** ******** ******.
Outperforms ******** ********** ** *** *****
** **** *** ** ** **** door ***** **** ** ******* *** of ~**** *** ** ****** *** of ~**.
*** ****** ***-****** ******** **** ****** of *** ******* *** **** ******* of *** ***** ****** *** ******* than *** ***-*****.
** ******, *** ***-****** ******** ******* to ***** ***** ******* *** ******** more *******. *******, ****** ******* ****** shadowed.
WiseNR ******** ** *** *****
****** ******* *** ***** ** *** scene, ****** ** ** ** *** expense ** ********* ********. ******* ****** details *** ******** ** **** *** QNV-C8011R *** ***-*****.
****** *** *** ******* ** *** questions ********* ****** ***********, *** ** will ****** *** ****** **/**** **** do.
VMS ***********
****** *-****** ********* ********** **** **** out ** **** ***** ***** ******, including ***********, *******, ******* *****, *** Milestone. ***** *** ** *********** **** Avigilon ******* ****** *** ***** ******.
Versions ****
*** ********* ******** **** **** ****** testing.
- ****** ***-******: *.**.**
- ****** ***-******: *.**.**
- ****** ***-*****: *.**.**
***, * ****** **** *** ***** Event ************ *** * **** ** error ****** *******. *** ****** **** were **********, * ***** ** ****** events ** ******** *** *** ****** that ********* **** ****** ****** ****** from *** ****** ****** *** *** object ** *** ******, ***** * could *** *** ** *********.
* ******* *** ***** ** *******.
** * **** ****, *** *** integrate ****** ****** **** ****** **** person *** ******* *********.
*** ********* **** ** *** *** Q ** ****** ** *** **** shot ** *** * ****** ** a **** ******* ** *** ******* in ****. ********* ****** ******* ******** with * * ****** ****** ** excellent. **** **** *** * *** more ****** ************* **** ** * quick ****** *******. ***** **** **** a ****** ***** ** *********.
**** ***** *** ** ******** ******** before ******* *****? ** * **** in**** ******,*** ********** *** **** *** **** IPVMs *******. ** **** * ****** QNV-C9011R.
*** *** **** **** ******** ***** Event *************?