The Risk of Reducing Sensitivity To Solve "Weapons Detection" Problems
"Weapons detectors," despite their branding, struggle to differentiate weapons from non-weapons. One solution is to decrease the sensitivity setting, but what are the security implications of this?
In this report, using difficulties experienced by one of Florida's largest counties and their adjustments as an example, we examine the structural pros and cons of lowering the sensitivity setting.
*** *******,******* ****** *** ***,***+ ******** ********, ****** ** *** ** *** biggest ********* ** *** *******. ******* schools *** ****'* ********, ***, ******* ***** ********, *** ****** ******** ** ***** CEIA's ***** ** *** **** ******* with * ******** ******** $*.* *******.
**** **** *** ******** ************ ** lower *********** ***** *** **** ** Openagte *** ** ***** ******* ********* as ****, ********* ****** **** ***** Express.
Long ***** ** *** ***** ***
** *** ****** **** ******* **** month, ******* ******* ****** *** ****** have **** ***** **** **** ******** late ** *******. *** *******, *******'* lines ****** **** ***** ********, **** multiple **** ******** ******* **** ***** (*,*):
*********, **** ******** ****** ** ****** and ********* ***** ********** **** **** lines (*,*,*,*,*). *** *******, *** *****:
Changes **** ***** *** ***** ***
***** *** ***** ***, *** ******* school ******** **** **** ***********, ********* reducing *** *********** *** ********* **** units ** **** *******, ** *** reported ** ******** ***** **** ******** (*,*).
******, **** ********* **** * ***** ***********, ***** *********** *************** **** *** *********** ******* *** changed:
* ******* **** ***sensitivity ****** might have been set a bit higher than it was supposed to. There was a *************. [******** *****]
*********, *** **** **************:
** ******* *** ********, *** ** the *********** **** *** *** *** involvedrecalibrating ***** ********* ***** *********** *** *** *** ****, causing them to be triggered by items like eyeglass cases and 3-ring binders. [emphasis added]
******, ******** *********** *** ** **********, students ********* ****** **** ***** ** passing "******* *********" ** *** ****** day ** ******, ***** ***** ** achieve ****** *********.
********* ** ******** *** *********** ** highly ******** *** ****** *********, ** often ******* ******** ** ******* ** the ********, ****** ******** ********.
Security ************
***** ******** *** *********** ******* *** reduce *** ***** ** ***** **** be ***** ***** ******, ** ******* significant ***** ** **** ******* **** be ******. *** *******, *** *** Evolv *** **** ******** ** *** to * ***** *********** *******:
******** ******** ******* **** ** ****** False ******,*****'* ****** ******** *********** ******* *** The **** ** ******* ********** **** ******* ** *********** ********.
************, **** ** *** * **** or ** *****-******** *****, ** *** commercially ********* ******* ********* **** * fundamental *****-*** ******* ****** ***** ***** alarms *** ******* *******. *******, **** has ****** **** ** * ******* as ********* **** ******* ********* **** "metal" ** "*******" *********" ** **** increases ********* ***** *** **** **** these ******* *** **.
Board ****** **** ** ******* ** ****** ******
********** ** **** *** *** *** new *********** ******** ***, ** ******** the ***********, ******* ******* ****** **** secure. * ******* ****** ***** ******, however,*********** **** **** ***** **** ** changes ** ****** *********** "********* *** *********** [***********]:
* ********** *** *** ******* *** reached *** ** *** ****** ***** members *** *** ************** **** *** question,"By ********* *** ***********, *** ** *** * ********** ** *** ****** *** ** *** *******?" I want to be very clear - **[we **** *** **** *** *******]. [emphasis added]
*** ***** ****** ********* *** ******* by ********* ** ******* ** * detected ********* ** *********:
****'* **** * *** **** ***, Glenna, ***** *** * ****** ***** we *** * ***** ********* **** actually *** ********* **** *** ******** by *** ****** ********* ******.
**** ** *** * ****** ******** because ************ ********* *** ******* ****** does *** **** **** ******** **** detect *******, ** ***** *** *********. To ****** *** *** ******** ******* with ***** *********** *** ** ***** has ** **** ******** **** *** same ****** ******* ** ********* ******** and **** ******* *** ********* *****.
Lowering *********** **. ********** ***** *********
******** ** ****** *******, **** ***** showed **************** ***** ********* *** ******* ********* to ******* *********, ** ***** ********* **** ******* sensitivity ********. ********* ***** ******** *** significantly ****** *** ****** ** ***** alarms, ****** ******* * ************* **** expensive *********** ** ******* *********:
********* *** ********** **** *** ******* Metal ********* ********* **** *******.
Longer ***** *** ******* *** *** ***********
** ****'* **********, ***** ************* ** "weapons *********" ** ****** ***** **** lines *** *** ***** *** **** (for *******, ****** **** ***** ** Atlanta **** *******, ***** ** ******* in******** ***** ****** ********* *****).
****** ***** *** * ****** ****** of *** *********** *********** ** ************ available ******* ********* ** ***** ********* to ************* ******* ****** ****** ******** objects *** *******. **** ****** ******* to *** ***** ******** ** ****** some ** ***** ********** *** **** them ****** *** ******. ** ******** have ******* ******** ** ***** *******, it *** ****** ******.
*******, *** ***** ** ******* ****** are ** *** *** ******* **** has **** **** *** ******* ******** deployment ** *-**.
Shorter ***** ** ****** ***
*** *********** **** **** **** ** Broward ******* ***** *** ***** *** (including *********** ******* *******) *** ** shorter *****, ** *** ******** ** several ***** ******* (*,*):
Tested ****** ****** ******
*******'* ******* ** *** ***** *** of ****** ** ************ ***********, ********** ***** **** ********* ** ******* ****** *******. **** ****** ** deficiencies ** *** ******* *** *** testing, ***** *** ****** ******** *** ~2 ****** ****** *** ***** *** of ****** ** ***** **** ***** the ******'* *********** **************.
"Miscommunications" **** ***********
**** **** ****, * ****** ***** member **** **** "****************" ******* *** school ******** *** *********** ** *** of *** ******* *** *** ************ experienced ** *** ***** *** ****. This **************** *** ** ******** ***** set ** * **** ********* ******* than ** "*** ******** **":
[*] **** ********** ******* *** *** and *** *** **** **** ** better?
[*] **'** *** **. ** *** clear **********. ** *** ***** * contractor *****. ***** *** *miscommunication between the district staff and the contractors ... I believe that the *************** ******might **** **** *** * *** ****** **** ** *** ******** ** [emphasis added]
Director ** ****** *** ******** ********
*******'* ******* ******** ** ****** *** Security ******** ** ****** **** ****, only ******* **** ***** *** ***** school ***, ******* ***** ********.
*** ****** ******** *** *** ******* the *********** ******** *** ****** ** interview **** *** ******, * ***** member *** *** ********* ** *** details ** *** ***********; *******, ** is ******** *** ****** ************** ********* to ****** ***** ***** *** ****** year *** *******.
Not ******** ** ******* ***** **********
***** ********** *************** ******** ** ****** *****-**** ******* (amongst ***** ****** ****** *******) *** pass **** ****** *** ****** ** an ******* ** *** ********* ***** the ***** **** ** ******, *** superintendent **** **** *** ** ****** needs ** ****** *******:
********, ****** ******** ** ****** **** laptops *****, *********, ****** *** ***** equipment. ****** *** ******** *** ***** detector,you ***'* **** ** ****** your cell phones, keys, ********* ***** ******* [******** *****]
** ****, **** ******* ******** ** it *** **** **** *** *********** is ************* *****, ********** *** **** that **** ******* *** ***** ******.
*********** ********** ** ***** *********:
*** ***** ********* * **** ****** told ****** "*** *** ****** ** false ******, ** ****** ******, ** something ** *** ******". ***** ** no **** ***** ** "** ***** alarms"
*** ***** ***** ** ***** ********* (that * **** *** **** *** for ******) ** **** ***** ********* did ************* ******* ** **** ******** / **** ****** ****** ********* ** the **** * *****, ***** *** not ***** *** ** ****** ******** on * ******* ***** *** *** better ** **** ******** ***** ********* and ***** *********.
********* **** **** *** **** **** occur *** *** *** ***** ***** of ****-******* ***** ********* **** **** new ************* ******** **** *** ****** differentiate ******* *** ****'* *** *** state ** *** *** ** ********** today.
***** *** *** ***** *** ** simply ******** ** * ******* *****
***, ***** ********* *** ***** *******. I ***** **** **** ******** **********, as ** *** ** *******, **** become *** ****, *** **** ******* improved ******* ***********. *** ***** *** still ** ***** ***** ****** ******** on *** ******. ***** ***** **** ultimately ****** ******* *** ******* *** events, *** ******* **** ***** *********, or **** ******* **** ******, ***** is ***** ** ***** *** ** miss **** ********** ** ******* **********.
*** ***** ******* ***** **** ** helping *** ***** *********, ************ *** the ****** ******** ******, ** **** really *** **** ****** ** **** about ** * ******. *** ** helps **** ** ******* *********, ****** tend ** **** ****** *********** ** the ***** *****. *** *** ****** figures *** * *** ***** ********** (ground *****, *******, ***** *****), *** can ** * *** ** ************* parse *** ******* **** *** *** big, *** *****, *** "******" (**: not ****** ***** *** ****** ***** or *******), ***. **** **** *** to * ***** **** **** **** objects ** ********, ******, **** ******* a ********** ****** *** ** **********.
***, ******** ****** ********* ** ****** detection. *** *****, **** ** ** start **** ***** ************ * ****** detector, *** **** ******* ****** **** person *** ******* ***********, *** ***** have * *** ** ***********. ******** and ******* *** ************ ********* ** size, ********** ********. **** *** **** similar ** ********** ** ***** ****** objects **** *****, *********, *****, *** things ******** ******* ** ******. *** might *** **** * **** ******* of *** ******, ** *** ****** thing.
****-******* ********* ****** **** ** *** video-based ******, *** *** *** *** relying ** ****** ****** **** **** classifier **** ** *** ******. **'* much ****** ** ***** **, *** you're ***** ******* **** *** **** that ******* **** ****** ********* ****** to ****** **** ******. *** ***'* screen ** **** *** ***** ** simple ********** **** ****, ** ******** patterns.
*'* *** **** ** *** **** of ***** *** ******* ********* ********** will *** ** ** ** **** as ***** ********* ******* **** (*** who **** ******* **'* ** ***** "video *********" ** * ******** *********** for ******** ***********??). *** ******* ********** all **** ****** **** *** ****, but ******* ******* ** ** ******** as *** ***** ******** ********** ** build ** *** ***.
***** ***** **** ********** ****** ******* not ******* *** ******, *** ******* some ***** *********
*** ******* **** *** **** ****. It's ************ *********** *** ***** ******** systems ** ****** ****** **** **%+ accuracy, ** ******** ** *** ****** a ***** ******** (*.*., ******, *******, etc.) ** ********* *** *** ****** it ****** ****** ****** (*.*., ***** negative).
*** ***** *** ***** ** ***** alarm ****** ******** ** *** ******
****, *********, ******** ********, ***** *** probably -*** *** ******** **** ***** have *** ***** ********* ** ********* but ***** ********* *** ****** ******, approaching **********, ***** "******* *********" ** in *** *** *** (*** ******, what * **** **** ** ** compare ** ** *** ********* ***** of ***** ********* **** ** **** simple *****-***** ****** ********* *** *** terribly **********). **** ** ***** **, I ** ***** ******* ********* = pixel-based *** ** * **** ******* to *** ******* **** ** ******* detection.
* ***** **** ******* ********* *** not ** ******* ****** ** ***** the ******* ** ****-***** *********. * still ***, *** **** **, ****** false ***** ****** **** * ******* of ******* *********, **** * ******'* say ** ** ******* **********.
** ******* *********, ** ********* ***** you **** **** ***** ***** ** persons ** *** ***** *** *** get ** **** **** **** ***********. The ***** ****** *** ********* ******* to ***** **** **** * *** of ***** **** ********, *** ****** with **** ********** (*** ****, ************ sites, ****** ******* ****), *** **** to **** ************* ****** ** ***** to ***** ****** **** **** *** in *** ********, ****** **** ******* or **** ********. **** ***** ** increased ***** ******. ** **** ******* these **** *** *** ***** ********* sensitives *** ********* ***** ** ***, but *** ******* ***** ** **** when *** **** ** *** **** animal ******** *** ***** ****** **** tend ** ******** **** ** ***** alarms **** *** *** ** ****** sensitivity ****.
****'* * ****** ******* ** *** video ********* ** *** **** ******* -******* **** ********, *********, *** ***** Detection ********* *************** *** ******* **** ****** ** the **** ****** *** *** ****** what "**********" *****. **'* ***** ****** in * ********* ***** **** *** problems **** *****'* *******/***** ********* ****.
*** ***** ******* ***** **** ***** analytics ** ***** ******* ***** ** still * ******** ******, *** ****'* more ** * ****** ****** *****, e.g. *** **** ** * *** or *** **** *** ***** *********** than *** ********* **********.
** **** ******** ** ******* ****** with *** **** **** ***** ********* are ***** ******** ***** ** ** problematic *********** ** *******/***** *********.
** **** ******** ** ******* ****** with *** **** **** ***** ********* are ***** ******** ***** ** ** problematic *********** ** *******/***** *********.
* ***'* ***** ** ******* **** made **** **********, *** ** ** did **** ****'* ** ******.
** **** ***** ** **** ** this ***** ***** *** ** ******* sensors, *** ***** ****** ***** **** be. ******** ******* *** ****** **** make ********* **** ****** **** **** cases.
** ***** ** ********** ** ** that *** ******** ****** ** ******* all ******* *** ** *******. ****** like * *********** ****** ** *********** happened ***** ****** ***** ******* - and ************ *** **** **** **!
**** **** ******** ** *** ******** space, ********** **** *** - ** pays ** *** *** ***** ************. What *** ******, ** ** *******, was **** * ****’* *** *** divestment ** * ********..
**** ******.
**** ******* ** ****** *** ******. For ***** ********* * **** ****** told ****** "*** *** ****** ** false ******, ** ****** ******, ** something ** *** ******". ***** ** no **** ***** ** "** ***** alarms", *** ***** ** **** ** such ***** ** "** ****** **********".
**** ****** ** ****** ** *********** (eg: **% ********) ** *********** ** best, ******* ***** ************ **** **** its *** ****** *******.