Amazon Rekognition Facial Detection Tested

By: IPVM Team, Published on Dec 03, 2018

Amazon's Rekognition service has announced improvements, that if accurate, would be a significant advance over conventional facial detection.

amazon recognition

In particular, Amazon claims:

Amazon Rekognition can now detect 40 percent more faces – that would have been previously missed – in images that have some of the most challenging conditions described earlier... These aspects might include pose variations caused by head movement and/or camera movements, occlusion due to foreground or background objects (such as faces covered by hats, hair, or hands of another person in the foreground), illumination variations (such as low contrast and shadows), bright lighting that leads to washed out faces, low quality and resolution that leads to noisy and blurry faces, and distortion from cameras and lenses themselves. [emphasis added]

IPVM developed a dataset of 100+ 3-second video clips:

facial detection dataset 2

These clips are done across a number of common real-world scenarios:

  • Day light (200 lux)
  • Low light (1 lux)
  • Side tilt (from 0° to 90°)
  • Down tilt (from 0° to 30°)

We compiled various combinations (such as 200 lux, 15° side tilt, 22° down tilt or 1 lux, 75° side tilt, 30° down tilt, etc.).

We then ran tests of these 100+ clips across facial detection approaches:

  • HAAR 
  • HOG
  • Openvino (specifically the face-detection-retail-0004 model on Intel i7 and Myriad X chips expanded upon from our NCS2 test)
  • Amazon Rekognition

Inside, we explain full results and key variances in performance.

******'* *********** ******* *** ********* improvements, **** ** ********, would ** * *********** advance **** ************ ****** detection.

amazon recognition

** **********,****** ******:

****** *********** *** *** detect ** ******* **** faces – **** ***** have **** ********** ****** – ** ****** **** have **** ** *** most *********** ********** ********* earlier... These ******* ***** *******pose ********** caused by head movement and/or camera movements, ************ ** ********** ** background ******* (**** ** faces ******* ** ****, hair, ** ***** ** another ****** ** *** foreground),illumination ********** (such as low contrast and shadows), bright lighting that leads to washed out faces, low ******* and resolution that leads to noisy *** ****** *****, and distortion from cameras and lenses themselves. [emphasis added]

**** ********* * ******* of ***+ *-****** ***** clips:

facial detection dataset 2

***** ***** *** **** across * ****** ** common ****-***** *********:

  • *** ***** (*** ***)
  • *** ***** (* ***)
  • **** **** (**** *° ** 90°)
  • **** **** (**** *° to **°)

** ******** ******* ************ (such ** *** ***, 15° **** ****, **° down **** ** * ***, 75° **** ****, **° down ****, ***.).

** **** *** ***** of ***** ***+ ***** across ****** ********* **********:

  • **** 
  • ***
  • ******** (************ *** ****-*********-******-**** ***** ** Intel ** *** ****** X ***** ******** ******** *** **** ****)
  • ****** ***********

******, ** ******* **** results *** *** ********* in ***********.

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

Key ********

****** *********** ****** ********* ******** was *** ****** **** all ****** ****** ********* nearly ** **** ** Openvino:

facial detection overall accuracy

*** ****** *** ********* quite '***' *** **** of *** ******** ** used **** ***** ****, including *** ********* *** bad ******.

*** *******, **** ** segment *** ****** ** key ********** ****, **** for ****** ***********, ***** differences ** *********** ******* are *******:

rekognition accuracy segmented

**** ******* *** *** severe ****** ******** ***, with '****' ***** *********** at ****** **% ***** low ***** *********** ** under **%.

***** *********** *** ********* in ****** ********* ********, it ***, *** *** away, *** *******, ** two ****** ** ********* versus ******** *** **** ******* on ** **: 

facial detection fps

*********** ** **** ****** **** the ***** ******** ** the *********** ***** ******* ******** the ***** ** ** sent ***** ** *** while *** ***** ******** can ** **** ******* / ** *** **** / ***** *** ****** is. ** ******* ***** measurement ******* ** *** final ******* ** **** report.

********

*****, ** ******* ** 100% ******** *** ****** Rekognition, ***** ** ****** strong *********** *** **** is **° **** *** 45° **** **** **** the ******:

[video-to-gif output image]

****, ** ******* ** 59% ******** (******* ~* out ** ** ****** correctly ******* * ****) for ****** ***********, ****** a ***** ********* ** steep **** *** **** tilt **** *** *******'* head ****:

*** ***** ******, ** noted, **** **** **** challenging **** *** ***********. For *******, **** * out ** ***** * frames *** * **** accurately ******** ****** **** a ***** ***** *** the ****** / ********:

 

*******, *** ** *** few **** ************ *** Rekognition *** ** **** test ********, ****** *** light, *** ******* ****** straight ******* *** ****** but ***** **** *** not ******** ** ***:

Accuracy ***********

*** ******** ****** **** is ************** *** * simplification. *********** **** ********* evaluation **** * ****** called"**** ******* *********" (***), ***** ******** ***********-**-******* on ****** **** ********* datasets ********* ***** ** ****** **** bounding-box ******* (********* ****** "Intersection **** *****" ** IoU) *** ******* **** box ********* ******** * face. ** ****'* ***** library, **** **** ****** has *** **** (***** frame *** * ****, and ***** ** **** one). ***** ****, ** can *** * ****** heuristic *** ********: ** how **** ****** *** a **** ******** (******** / *****-***** = ********). This ** *** *******: in **********, ** **** not ******* *** ***** positives (* **** ** detected ***********, **** ** the **** ** ** the *******'* *****). **** happens ****** - ** we ***** **** *** heuristic. ** ****** ** more ******* **** ********** (lux, *****, ***) ******* having ** ***** ******** boxes *** ***** ***** of *** *******.

FPS ***********

*** ******* **** ****** used *** *********** ***:

  1. ****** *** ***** ** Amazon **
  2. ***** *** *****
  3. ***** *********** (******** ** ** the ** ****, *********** an***->***********)
  4. **** *** *****, ********* FPS

* **** ****** ******** for ************ ***** ** to ****** ****** ******** via *******->***********. **** ***** *** the ********** ******** ** a *****-**-***** *****, ***** our ***** ***** *** networking ******** **** *** FPS *********** ********. ** such, *** *** ***********, already **** ***, ***** be **** *****.

***** ** **** ************, this ** * ***** point ******* ***********, ** others ** *** ** a *******. *** *******, one *** *** ***-***** hardware *** ****** ******* performance. *** ***** ****** in **** ******* *** on ****** * ** Intel ** ******* (****-***) devices, ** **** *********** trade-off ** *********-*********. ************, Rekognition ** * ******* service, ***** *********** * model (**** ********'* ****-******-****) takes **** ************* *** setup. ** ***** ****-**** performance **** *********** ** ******** (note: ** *** ***** investigating ** **** ** possible), *** ***-****-**** ************, Rekognition ***** ** ***** useful.

*******

*** * ****-**** ***** surveillance ***********,*********** ************ ** ***** **** but *** ***** ************ where **** ********** ***** processing ** ******, ** could ** *** **** affordable **** **-**** **********. For *******, *********** ******* for **-**** (*. ********) is $0.12 *** * *** of **** ****** ***** analyzed. **** **** **** 10% ******** ** ******* (i.e., *,*** ******* ** a *****), *** ******* **** would ** ~$** (*,*** x **% * $*.**) just *** *** **** analytics.

** *** ***** ****,****** *********** ****** ******* ***************** ****** ***********, ****** analysis (***, ******, *******), object *** ********** *********, etc. **** **** ********* provider ****** ** *** have ** ***** ** likely ** ******** ** compete **** ***** ***********.

********

**** ** *** *** of ****'* *** ****** of **** ******** *****, starting **** ****** ********* ******, ***** ****** ******* ***** 2 / ******** ** Test *** *** ****. ** are ******** * ******* of *** ***** ** the ******** ******. *** feedback, ******** ** *********, please *** ** *** comments.

Comments (7)

John Honovich

IPVM | 12/03/18 01:52pm

*** ***** ** *** interested ** *** ************,***** *** *** *** class **** ****:

deep learning webinar2

***** **** ** ********** fundamentals ** **** ** our ******** ****** ** provide ************* ******** *** education, ** ******** ** our ******* *****.

Undisclosed Manufacturer #1

12/03/18 02:50pm

*** *** ******* **** running *** ********** **** IR *********** **? **** cameras ***** ****** ** IR *********** (** **** have *** ******) ** 1 ***

John Honovich

IPVM | In reply to Undisclosed Manufacturer #1 | 12/03/18 09:02pm

** *** *** ** any ** ******** *** we **** *** **** in ****** *****.

Avatar

Matthew Del Salto

BLACKBOX USA | 12/03/18 03:12pm

**** **** ***** **** integrate ** **** **** if **** *****'* *******.

Undisclosed #2

12/03/18 07:45pm

** ***** *** *** you *** **** *** of **** *** ***** used ** *** ***** for **** ***** ** download *** *** **** their *** ********* *******?

Undisclosed Manufacturer #3

In reply to Undisclosed #2 | 12/03/18 08:02pm

* ***** **** ***** be * ****** ***** way ** ******* ********.

John Honovich

IPVM | In reply to Undisclosed #2 | 12/03/18 09:03pm

** ***** ******* *** dataset *** **** *** decided ***.

Login to read this IPVM report.
Why do I need to log in?
IPVM conducts unique testing and research funded by member's payments enabling us to offer the most independent, accurate and in-depth information.

Related Reports on Low Light

Dahua 4K Camera Shootout on Aug 20, 2019
Dahua's new Pro Series 4K N85CL5Z claims to "deliver superior images in all lighting and environmental conditions", but how does this compare to...
Axis 4K Camera Shootout 2019 on Aug 14, 2019
Axis' 4K Q3518-LVE claims the "best video quality possible", with Lightfinder super low light performance, Axis' high end Forensic WDR, and...
Hikvision 4K Camera Shootout on Aug 02, 2019
With their latest Smart Series 5 cameras, Hikvision is claiming cameras "fully loaded" with "state-of-the-art technology for high performance and...
Hanwha 4K Wisenet P Camera Tested on Jul 30, 2019
Hanwha claims "premium performance" and to "deliver the best in high resolution performance and efficiency" with their Wisenet P series 4K models,...
Bosch Starlight 8000i Cameras Tested on Jul 23, 2019
Bosch has released their new Flexidome IP Starlight 8000i cameras, claiming "exceptional detail even in extreme low-light situations." To see...
History of Video Surveillance on Jul 19, 2019
The video surveillance market has changed significantly since 2000, going from VCRs to ab emerging AI cloud era.  The goal of this history is to...
Axis ARTPEC-7 P1375-E Camera Tested on Jul 12, 2019
Axis claims the new P1375-E box camera with ARTPEC-7 chip delivers "clear, sharp images in any lighting condition." But how well does it do? We...
Last Chance - Camera Course Summer 2019 on Jul 11, 2019
Last day to register is Thursday, July 11, 2019. This is the only independent surveillance camera course, based on in-depth product and technology...
Hikvision ColorVu Camera Tested on Jun 24, 2019
Hikvision says their new ColorVu line captures "vivid chromatic images in darkness", with unconventional white light illuminations whereas most...
Repositionable Multi-Imager Camera Shootout - Avigilon, Axis, Dahua, Hanwha, Hikvision, Panasonic, Vivotek on Jun 19, 2019
Repositionable multi-imager cameras are one of the fastest growing segments in video surveillance, with a slew of new offerings being recently...

Most Recent Industry Reports

Dahua 4K Camera Shootout on Aug 20, 2019
Dahua's new Pro Series 4K N85CL5Z claims to "deliver superior images in all lighting and environmental conditions", but how does this compare to...
ZK Teco Atlas Access Control Tested on Aug 20, 2019
Who needs access specialists? China-based ZKTeco claims its newest access panel 'makes it very easy for anyone to learn and install access control...
Uniview Beats Intel In Trademark Lawsuit on Aug 19, 2019
Uniview has won a long-running trademark lawsuit brought by Intel, with Beijing's highest court reversing an earlier Intel win, centered on...
Verkada People And Face Analytics Tested on Aug 16, 2019
This week, Verkada released "People Analytics", including face analytics that they describe is a "game-changing feature" that "pushes the...
Dahua OEM Directory 2019 on Aug 16, 2019
US Government banned Dahua OEMs for dozens of companies. The following directory includes 40+ of those companies with a graphic and links to...
Installation Course - Register Now on Aug 15, 2019
Register Now for the September 2019 Video Surveillance Install Course. This is a unique installation course in a market where little practical...
Axis Suffers Outage, Provides Postmortem on Aug 15, 2019
This week, Axis suffered an outage impacting their website and cloud services. Inside this note, we examined what happened, what was impacted...
Hikvision Scrutinized In The Netherlands on Aug 15, 2019
Hikvision is facing unprecedented scrutiny in the Netherlands, at the same time the US government ban has taken effect. This week, a Dutch...
Axis 4K Camera Shootout 2019 on Aug 14, 2019
Axis' 4K Q3518-LVE claims the "best video quality possible", with Lightfinder super low light performance, Axis' high end Forensic WDR, and...

The world's leading video surveillance information source, IPVM provides the best reporting, testing and training for 10,000+ members globally. Dedicated to independent and objective information, we uniquely refuse any and all advertisements, sponsorship and consulting from manufacturers.

About | FAQ | Contact