Google’s Super Resolution Enhancement Examined

Author: Brian Karas, Published on Feb 10, 2017

"Zoom in and enhance, I think there is a clear shot of his face in the reflection" has been the kind of statement CCTV users have wanted to make reality for years.

Hollywood has imagined this technology, but now engineers at Google have released a whitepaper on "Pixel Recursive Super Resolution", showing real-world examples of upsampling a highly pixelated image into one with details and recognizable features.

In this report we examine Google's super resolution whitepaper, and how this technology could impact the security industry.

"**** ** *** *******, * ***** ***** ** * ***** shot ** *** **** ** *** **********" *** **** *** kind ** ********* **** ***** **** ****** ** **** ******* for *****.

********* *** ******** **** **********, *** *** ********* ** ****** have ******** *********** ** "***** ********* ***** **********", ******* ****-***** ******** ** ********** * ****** ********* ***** into *** **** ******* *** ************ ********.

** **** ****** ** ******* ******'* ***** ********** **********, *** how **** ********** ***** ****** *** ******** ********.

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

Recursive ***** ********** ********

*** ********** ********* ** ********* **** *** **** *** ********** input ******, *** ************ *********** ** ****** **** ********* ******/**********. When ******** ** *** ****** ***** *** ***-********** ***** *** derived ****, *** *********** ****** **** **** ************, ****** ** that * ***** ******** ***** ****** ********* *** ******, ***** they ***** *** **** **** **** ** **** *** ***-*** image.

** ******* ** *** *********** ****** ***** **** *** *********, its ******, *** *** ****** ****** ***** *** ***** *****:

*** ******* ** ********* ** ***** ******* ** ** ****** creating * ******** ** * **** **** * ******, ** from * **********, ***** *** ****** **** ******* ** *** painting ***** ** ** ************* ** *** *** ******* ****** be ******* **** ***** ************ ** *** *******.

***** *******/****** *** ** *************, ** ** *** ******* **** to *****, *** ********** **** ***** ******** ** ****** ** bedrooms **** **** ************* **** ******* ***-*** ***** ******.

Requires ***** ***** ******/********

** ***** ** ******* ******** ******, *** ********* ***** ** be ******* ********** ** * ****** ** ****** ********** ******. If *** * ***-*** ***** **** **** *** ********* ** people/objects ** *** ********** ******* *** ******* ****, *** ******* will ****** ** ****** **********.

** ********** *** ***** ** *** **** *** ******* ******** process *****, ****** *** ********** **** ***** **** * **** were **** ** *** ******** *******, ********** **** ** ** very ******** *********.

Limitations ** ***** ********** ****** *** ********

******* *** ****** ****** *** *********** ********-********* *******, *** ****** would ****** ** ************ *** ******** ** ********. ******* ** police ********, * ****-***** *********** ** **** ********** ** ******** might **** ****** **** * ******, *** ***** ****** *** be ****** ** ******** ** *** **** *** * **** high-resolution ********* ***** **.

Compared ** ****** ***********

****** **** ********* ****** ****** ** *** ******** ***, *** this ** *** ******** ** ** * ****** *********** *******, and *** ************ ********* ******* ********* ***-*** ****** ** **** them **** *****-********* ** ****** ****** ***** ** ******** ******** from ******* ******. *** ********* ** *** ********** ** *** details ******* *** ******* ********** ******, ** ** ********** ** create ******* ***** **** ***** ** ** ***** *****.

Applications ** ******** ********

**** ********** ** ******** ** ***** *********-***** "**** ** *** enhance" ************* ** *** ****** ******* ** **** *****. *** reliance ** ***** ******** ***** **** ** ******** *** ********* for ********, ** ********* *** ***** ******* *** ******* ** persons ** ******* **** **** *** ****** **** **** ** the ******* **********. ***** ***** ** ************ ** *** ****-*** images ** ******* **** *** ****** ** *********** ******** * person ** ****** ** *** ********** ** *** ***** **** another ******, ****** *** ******** ** ** ************ ********** ****** and ******** *** ********* ***** ****** **** **** *********** ***** the ******** ************ *** ************* *******.

Comments (21)

Clayton Burnett

02/10/17 03:23pm

**** ********* *** **** ***** ******* *** ****...**** ** ** gave ** * ****** ***** ** **** *** **, ** could ** **** ********** ** **/*** ******.

Thumbnail brk1

Brian Karas

IPVM | In reply to Clayton Burnett | 02/10/17 03:49pm

**** ** ******** ********* **** *** ****, ****** * *** of *** **** ******** ********** *** *** ****.

** ***, *** *** ****** ** **** ** ***** ***** for * *** ** **********, **** ***** ***** ** *** problem ****** ******* *** *** ********* ******* ** * **** small *** ** ************* (*-*, *-*, ***.). ****, ** *** can ****** *** (** *****) *** ***** *** *** ** able ** ***** ** ******* ** *** ***** **** ********** sequence *******.

*** **** ****** ***********, **** *** ****** ********** ** "*****" an ***** **** ***** ********** ** * *****. ** ** less ***** ******* ********, *** **** ***** ******* ** ******* that **** *** ******* ************. **, *** **** ******, ** would *** ** ***** *** ***, ** *** ********* ***** be **** ** **** * **** ** ****** *** **** it **** ** ***** **** ***** **** **** **** **** you ***** ****** * ******* ***** ** **** ****, ** would ******** ** **** ********* **** *************** ******* ** "*" and * "*" (*** *******). **, *** ******** **** ***** cause ** **** ***** ******- ** **** ** *** ***** plates *** "*"*, *** **** ** **** *** "*"*, ** might **** * ***** **** ** * ***** **** *** "QQQ" *** ***** ** ** "***". (**** ** ** ************** from ******* *** **********, ***** ***** ** **** ** **).

Undisclosed Integrator #2

In reply to Brian Karas | 02/10/17 04:28pm

********** ** **** **** *** --

** ** **** ****** ** ***** * ***** ********* ********* that ***** ******* *** *, *, ** * **** * wildcard **** ***** ****** ****** *** *. **, ** ***'** telling ** **** ** ***** **** * **** ** ****** and *********** ****** ** ********* **** ***** **** ** * range ** ******** *******, **** ***** ** ********* ******* ** a ****, ******* *************. ** ***'** **** ** **** *** police * **** ** **** ** *** ** *** ******* plate ******* *** **** ** ****** ******, *** ********* **** the ***** ** *** ***, **** *** ****** * **** down ********** *******.

*** ******* ** ****** * ********** ***** *** ** ********* totally ************** ***** ** ****-******** *** **** *** ******** *** investigations ******** *** *** *********** *** ****** **** *******. *** ability ** *** *********** ** **** *** **** ***** ** evidence -- **** ** ******** ***** -- *** ***** ** narrow *** ***** ** ********* **********. ***, ** **** *****, you ***** **** ****** ***** ******** ** ****** ** ********* data ** ******* *******, ** ** ***** ********** "*****" **** a ******** ** ***** ****** **** ***** *****, ***.

* **** ** ** ******* ****** ****** ****, *** ** far *** **** *********** ***** ********* ** *** ********** ******** industry ***** *** ** *** *********** ** "*** ****" ** data ********* ** ******* ****** ******** ********** ** ********** *********** to ***-*****. ***** *** **** ******** ****** ********* ** **** regard ******* -- * ***** *** **** *** ***** ** going ** *** ** ********* ** **, ************ ** *** 1 ** * ********* **** *** ********* ***** ** ***** to **** ** **** *** ****.

Thumbnail bischof 0773r burnbg island

Matt Bischof

02/10/17 04:23pm

**** ***********... *** **** ** *****-*************.

Undisclosed Manufacturer #1

02/10/17 04:24pm

**** *****, *** ****** **** * ***** ************* *********...

Undisclosed #4

In reply to Undisclosed Manufacturer #1 | 02/11/17 12:35am

*** ***** **** ***** ******** *** **** ** *****, * wonder?

** ******* ** ******* * ****...

****, ** *** ***** ******** *** **** ******** *** ***** to ******, **** * **** *** ***** ** **** ******* video ********.

Undisclosed Manufacturer #6

In reply to Undisclosed #4 | 02/13/17 05:21pm

******** * **** ******* ******** ***** ***, "*** *** *** focus ** ** ****** ** *** ***** *****? **** ** through *** ***** *** ****, ******."

Undisclosed #4

In reply to Undisclosed Manufacturer #6 | 02/13/17 06:59pm

*** *** *** ***** ** ** ****** ** *** ***** place? **** ** ******* *** ***** *** ****, ******.

"** **** *** ******, **** ******* ***** ******** *** *** it **** *** ****** ********. **** ******** **** ******* *********** about ******* ****** *************** ** ********* ** ***** ***** *** or *** *** **** ** *********** *** *******. ** **** case, *** ************ ******** ***** ******** ******* ********* ** ******* that *** ********** ****** **** *** *******. *** ******* *** investigated *** ***** ***** *******. *** ******* *** **** ***** to ** ** ********** ** *** ****** ***********."

Thumbnail bischof 0773r burnbg island

Matt Bischof

02/10/17 04:36pm

*********** *****'* ****** *********** ************ ** ****** * *** **, were * *** ** ********* **** ***** ********* (*** ******), parse ***** ******** *** ******* *****' ****** ****** **** *****/****** captured *** ************, ***'* ** ** ** *********.

* ****... ***'* ** *** ** *****'* *** ******* ****** to **** ** *****'* **-**** *** ********* **'** ** ***** to ****** ******* *** ***** '*****'?

Thumbnail brk1

Brian Karas

IPVM | In reply to Matt Bischof | 02/10/17 05:36pm

**** -

* ******* ***** *** ****** ***** *********** *****. **** **** on ********/***. **** ** ** **** **********, **** ***** **** a **** ****** ******* *** ***** ******.

*******, ***** *** ** ********** *** ** ********* **** **** that ** ****** ** "*****", *** ****** ** "*****". *** could *** ** **** ** ******** ** ****** ***** **** is *** *********** ** ****** ******** ** ******* ******. *** the ******** ** ******** * "********", *** ***** ***** ** highly ************** ** *** ******* (* ****** **** * ***** face, ***** **** *** ***** ****, ** * ****** **** a ****** ****, **** **** *** *** ****), *** ** might *** ******** ** ***** ** *********** * ******** ******.

** *** *** ***** * ******* *******, *** **** *** employees **** * ********, ** **** *** ******** ** * campus, *** **** ****** **** ******* **** ****** ***** *** specific ********** **** ** *** *** * ******* ** "*** people".

**** ** ***** ******** **** ** ******* ** ******, ** comes **** ** ******** *** **** ********** *** ** "**********" and ***** **** **** ** ****** *********. ** *** ********** they *** ** ******* ** ******** ****, ** *** ******* the ***** ****** ** *** **** ******** *** ***** **** up **** * **** **** "*******" ***** **** ***** ****** not ** ********** ** *** *******. *** *** ***** ******* minor *********** ***** **** ** ***** ** ***** ** ******* output ****** **** **** ***** **************.

***** ** ** *******, **** ********** ***** ** ****** ** create * **** ********** ***** *****. ** *** *** ******** the ********** ** * ***** **** ** "***** ** * red *****, * **** ******** *** * ***** ******** *******" that ***** ** **** **** ** ***** **** ********* *** actual ****** ** ***** *******. **** ******* **** *****, *** algorithm ***** ****** *************** ** ***** ******* ** *** ***. Or *** ***** ***** * *** ********* ****** ** *** vehicles, ***** **** ******** *****, *** **** *********** **** *** objects ** ***** *** ***, **** ******* **** ***** *********** to ****** ****** **** ************* ********* *** *** ******** ***** look ** **** ***** ** ******, **** ** *** *** not **** **** ****** **** ********.

*******, ** *** ********** *** ** **** **** ******, *** with ** "*******" *********, ** ***** ** **** ******-****, ******** the ****** ** ****** *** ****** **** *** ******* ****, while ***** ******** * ****** ** ***** ******** * **** detailed *****.

* ** *** ***** ** **** *** ************ ** **** in *** ****-**** *** ************, *** ******* ** **** ***** of ***** ********** *** **** ** **** ******* ************.

Undisclosed Integrator #5

In reply to Brian Karas | 02/13/17 02:44pm

***'* **** ** *** ** *******. ***** *** ** ***** of ***********. *** ** ** *** ******* *** ** *********** with ************, ** & ********* ** *** **** **** ******.

Undisclosed Integrator #3

02/10/17 05:02pm

** *** ***** ** **** ** ******** ** **** *** footage **** *** ****** ** **** *********? ******* ** **** is *** **** *** ***** *** **** ********* ** *** your ******* ** **** *** ***** ******* *** "*******" **** each ****** ** *** **** ***** ** ****** *** ****** plate.

Thumbnail brk1

Brian Karas

IPVM | In reply to Undisclosed Integrator #3 | 02/10/17 05:58pm

** *** ***** ** **** ** ******** ** **** *** footage **** *** ****** ** **** *********?

** ******, ***, *** ******** ** ** **** ************* ****. The ********** ********** ***** * ***'* ** ****** *** ***** from * ****** ** ****** ******. * *** *** *** an ***** ****** ** *** ******** ** ***** ******, *** even * ******* ******* ** * *** ***** ******** ** images *** ******, *********** ********* * *** ** *** ********** to *******.

**** **** * **** ****, *** ************* **** **** *** with ***** *****-********** *******, * ***** **** ** *** ** the ***** ** ********** ***** ****** *** **** ******** **** it ***** ** ******** ***** **********, *** ***** **** ********* in ***** ** ******* ************ (** **** ********* ** ***** of ******** ****). ** **** ****** **** ******* ****** ** generations ** *** ************ ****** **** ******* *** **** ** thing **** ** **** **** ** ** ******* ** **** end ******** ******.

Frantz Mathias

In reply to Brian Karas | 02/13/17 09:49am

**

**** ** *********** ...

*** *** **** **** ** ***** ** **** ** ***** processing ***** *** ******** ** **... ***** ********** ***** ***** to ** ******** **** ******** ** *** ***** ** ******** more *********. ** **** ** ********* * ****** ***** ** performed ** * ***** ** ******* ** *** *****. *** results ***** ** ***** ** **** ** * ***** ********** if ***** ** ...

***********? *******. * ******* **'** ******* *****.

Robert Sears

In reply to Brian Karas | 02/16/17 05:55pm

*** ** *** ***.. *** *** **** *** *** ***** federal ******** *** ******** $$ ****** ** ***** **** ****** than *********. *** *** *** ******* *** ******* **** ****** to *** **** **** *** **** ** ******, ** ***** for **, ** *** ***** ****. **** ******** *** ** "created" ****** *** ** **** ******.

Robert Sears

02/16/17 02:58pm

**** ** **** ***** *****.

* ******** ***** *** ** ** ****** ** ** ********* for ** ** ** ********** ** ***** ******** *** ** should ***** ****** ** ****** **** *** ***** ************ ***** really ** ***** ** **** ********* **** ** ****** **... if **** *** ** **** **** **** *** **** **** that ***** ****** ** ** ******** **** **** ** ** be *** ** **** *** **** *** ******** ***********.. ******** the *********** *********** ***** ** * **** *** ** ***** and ********* ** ***** ****** ** *** *** ****** ****. Think ** *** ***** "***** ** *** *****" ** *******.. No ****** ******.. ****** ****** ******* *** ******.

Undisclosed Manufacturer #6

In reply to Robert Sears | 02/16/17 04:36pm

* ***'* **** **** ******'* ******** ** ****** ******** ******** here.

******* *** *** * ********** ********* ********* (**** ****** ** from ****** ****), * ****-********* ********* (** ***** * ****** clicks * ****** ** ***** ** *********** *** *****'* *********** know **** *** ****** ****), ** * ********** ****** **********, you ***** **** *** ***** **** ** **** *** ******** image *** **** *** *** ******* ** *** ******** *****.

*** **** ******* ***** ***** ** ****** (** ****** ****) was **** ** ****** ** ******* ***** **** ***** *** be ********** ** *******.

* ** *** ** ***** ******, *** ** ******* **** this **** ****** **** *** **** ***********.

Undisclosed #4

In reply to Robert Sears | 02/16/17 05:17pm

...** **** *** ** **** **** **** *** **** **** that ***** ****** ** ** ******** **** **** ** ** be *** ** **** *** **** *** ******** ***********...

******* **** * ****** **** * ****** ******. **** **, both *** ** ****** ** **************.

Thumbnail p1010364

Tim Howell

IPVMU Certified | 02/16/17 06:08pm

***

*********** ********** *** * ** *********. **** ******** *** ********** better ******** ***** ** ** ****. *** * ** **** technology *** * ********** **** ** ****** ******** *** ********* processing...

*** ******* **** *** *** ********* ** ******* ****** ** those ***** ** *** ***** **** *** ******** ********. *** example *** **** ***** *** **** *** ******* **** ***** and *** **** *** ******* *** ***** ******** ** *** eye ****** ** **** **** ********. **** * **** ** the ****** *****, *** ********* *** *** *** *** ********* of *** **** ******* *** **** * **** ** *** bottom *****, *** **** ** *** ***** ** ******* ******* to **** *** ********* *****. ** ******* ** *** ******** between *** *** **** *** ***** ***** ** *** *** and ****** ********, **** ***** ** ** * ****** ******. The ****** ** *** **** *** ***** *** **** ********** between *** ********* *** ****** *****. **** * ****** ** see ** *** ***** ******** (** ** ***** *** *** two) ** ****** ********** ******* *** ****** *** **** ********** beyond. * ** *** *** ****, ********** ** *** *** image ***** *** ********** ****** *** *** **** ***** *****. Missing **** ***** ******* *** ********** ** *** **** *************.... The *** ***** ***** **** *** ********* ****** **** *** actual ***** ****** *** ********** ** ***** **** ******* ** age **** *** ********* *****. *** ***** ****** ** *** pictures ****** ** ** ******* ** ********* ********** ** *** algorithm **** ** *** ****** ****** ******** ** *********, *** lower ***** ** **** **** ** **** *** ******... * would **** ** **** **** ***** *** **** ********* *****, I *** ***** ***....

******* ***** ** **** *** ******* ** *** ********* *** be ******* ******** ** *** ******* ******** ********** ** *** picture (** ******* ** *** ******* ********** ** * *** resolution ******). ** ***** ********** ** **** ************ ****** *** quantization (* ** * **********) *******, ** ***** ** ** that ***** ***** *** ********* *** ** ****.

Rashid Khan

In reply to Tim Howell | 02/17/17 02:54am

**** **********, * **** **** **** ** ********** **** ********** in **** ***** ** **** ** **** *** (********** **** younger **** *********** ********), ** ***** ********* ** *********** ** run **** ****** ***** ********** ********* ******* ***** ** ******* origin *** ******* *** *******, ** *** ***'* *******, *** results ***** ******** *** ** **** **********. ****** **** ****, this ** ***** * **** *********** *** ********* ******** ** FR *** ** *** ***** *********, * *** *** * lot ** ********* *** **** **** ******** **** *** ****, will **** * ***** ** *** *****.

Henry Detmold

02/18/17 02:04am

** ***** **** *** **********:

** ** **** ***** ************** *****, *** ******* ******* ** super ********** ** ******************** ******* **
******** * **** ** *** ****** ************.
** ** ******** **** ************** ***** ** ******** ** ********!
***** ******* (**** **** ******** *** ******** ********) ****** * database ** **-********** ********* ****** ***** *** **** **** ** generate **-********** ****** "********" *** ********** *****. *** ******* ** might ** **** *** ****** *** **** *** * *********** of ********** ****** (*** ********* ** **** *******).
*** ******* **** ******* *** ********* **** * ***-*** ***** that ** ********** **** **********-****:
  • ** ******** **********, *** ****** ****** ** * **** ** average ** *** ******* ***********, *** ******** *** *** ** particular (* ********** *******)
  • *** ******* ** **** ******** ** **** ** **** **** a ********** ****** ** ***** *** ** *** ****** ** its ********, ****** **** * ***** / ******* (** ** will ** *** ** ******** ** ***** *** **** *** be * ******** ***** ** *** *****). ** **** *** seem ** ** ****** ******** **** ** **** **** *** correct **********, *******: ** **** **** * ********** ******, *** perhaps *** ***** ******.

***** ***** ** ** *********** ***********, *** ***** *** **** be ************ ** ************* ************, ***** * ******* ************ ***** output *** ********* ********** *** * ***** ***-*** *****. *** example ** ** ************ ******* ***** **** *** ****** ** an **** ** * ***** **** *** ******* ** * small *** ** ***** *********** ** *** ** **** ** use **** ** ***** ******* ************* (*** ********* **********) ** to ***** ** **** ***** *** ** ****** ** * given ***-*** *****.

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.

Most Recent Industry Reports

Final Day Save $50 - IP Networking Course September 2017 on Aug 17, 2017
Today, Thursday, August 17th is the last day to save $50 on the September IP Networking Course. This is the only networking course designed...
Directory Of Consumer Security Cameras on Aug 16, 2017
The consumer camera segment continues to grow, with new startups and models from existing players released seemingly every month. In this report we...
Cat 5e vs Cat 6 vs Cat 6a Network Cable Usage Statistics on Aug 16, 2017
Cat 5e? Cat 6? Cat 6a? What do integrators use in practice, today? 140+ integrators told IPVM. Here are the results: For those who want to...
Hikvision Responds To Cracked Security Codes on Aug 15, 2017
Hikvision has responded to IPVM's report on Hikvision's security code being cracked, both with a 2 page update to dealers and communication...
Stolen Video NVR / DVR Statistics on Aug 15, 2017
"But what happens if someone steals my recorder?" Anyone who has done more than a handful of jobs has probably heard this question several times....
Hikvision Europe Cutting Out Unauthorized End User Sales on Aug 15, 2017
The days of anyone buying Hikvision from anywhere off the Internet are numbered, at least in Europe, if Hikvision's plan comes to fruition. In...
Axis Laser Focus PTZ Tested on Aug 14, 2017
Axis has been touting its new Q6155-E laser focus PTZ as 'always in focus' and 'always in color'. Does it really deliver? We bought and tested...
Vulnerability Directory For Access Control Cards on Aug 14, 2017
Knowing which access credentials are insecure can be unclear, especially because most look and feel the same. Even the most insecure 125 kHz types...
IP Camera Specification / RFP Guide 2017 on Aug 14, 2017
RFPs are hard. Do them 'right' and it takes a lot of knowledge and time. Do them 'wrong' and you can be (a) unwittingly locked into a specific...
Cellphone Usage Issues For Integrators (Statistics) on Aug 11, 2017
Cellphones clearly offer significant advantages in communication and problem solving. But they can also be a major pain point if employees...

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