YOLO-Based Lilin Analytics Improvements Tested

** **** *******, *****'* ****-***** ********* performed **** ******. ***, ***** ***** us *** ****** ******** (*.*.*.**-******) *** improved ***********, *** *** ****?

***** ******* *** ******** ** ********* ********* ******** **** - ** Manufacturers, ** *********.

** **********-********** *****'* ****** ********, *********:

• *** *** *********** ******** ******** ** their **** ******** ******?
• *** **** **** ** ****** ****** sources ** ***** ******, **** ** foliage, *******, ** ********* *******?
• **** ***** **** ** *** **** cause ***** ******?
• **** ** *********** *** *******?
• *** ****** ** ******** ** ********* postures ** ********** ** *******?
• ** ***** ****** ***** ** ******* loitering ** *** *****?
• *** **** **** ** ***** ****** across *** *****?
• *** ****** ** ********* ** *** setup?
• **** ** ********* **** *** *** party ***?

*******

*** *******-***************** *** ** ********* *** $*** per ******* *** ******** ****** ** ****** ******, ***** *** **** ** ********* for ~$*** - $*** ******. *******, this ** **** ********* **** **** Axis' ****** **** ~$*** *-****** ****** with **** *********, ***** ** ******** with *** ****** ******** *** **** not ******* ** ********** *******.

*******

** *** *******, ***** ****-***** ********* showed ***** *********** ***********. ********, ** still ****** ***** ******* ******** ** average ********** **** *********, *****, *** Verkada. ***** ********* **** ** ****, foliage, *******, *** ********** **** *******. However, ** ******** *** ********* ******:

• ***** ********* ******** ** ****** ***** areas, ********** ** ******** *******.
• ******* **** ***** ************* ** ******, happening ** **** * *** *** night (**) *******.
• ****** ** *** ***** ***** ********* multiple ****** *** ** *********** ************ in ********** ****** ****** **************.
• ****** ** ****** ******** **** ************* reduced ******** ** ******** ******* *** still *******.
• * ****** **** ******* *** (~**-**) was ******** *** ******** ****** *********.

*****'* ********* ******* **** ******** ******* for *****-******* ***********, ** ***** *** seen ** **** **** *********. *** most ****** ***** ******* ********* **********, dwell **** ** ******* ** *** setting ** ****** ******** *******.

Minor *********** **** ******** ******** ***********

*****'* ****** ******** **** ******** ********* performance ** ************* ********** *** ****** of ******** ****** ******** ** * single ******, ****** ****** ******** ** the *****, **** ******** ****** ** ~1 ***** ****** *******. *******, ***** positives ** ****** ******** **** ***** present.

** ***** ************ **** ******** ****** testing. ***** ***** *** ****** **** random ***** ********* ** *** *****, inconsistent ****** ********, * ******-****-******* *** needed *** ****** ********* (~**-**), *** animals ********** ** ****** ** *** scene.

****: ** ***** **** ******* *** one ** *** ******* ****** *** AI *********, *** ** *** *** testing **** **********. *** **** **** saw ** ***** ********* ** ******* being ****** ** * ****** ******* through * *****, *** ** ***** that ** ****** ********* ** * scene **** ***** ***** *********.

Below ******* *********** ******** ** **** ******* **********

** ***** **** *****'* ********* ********* below ******* ******** ** ******* ********** like *********, *****, ** *******. *** average ********** **** ***** *** ** have ****** **** ***** ***** ******* such ** ****** ******** *** ****** objects *** ******** ******* ****** ****** the *****, ***** ***** *** ****** with *** *****. ************, *****'* ****** tracking ********** ***** ** *** ***** people ******* ******* *** *****, ***** the *** ********** ***** * ****** without ***** ****** *** ***** ****** their *** *****.

*******, *****'* ***** ** *********** ******** to **** **** *********, ********* ******** numerous ******** *** ******** ***** ** add ****** ** ***** *** ***.

Complicated ***** *******

***** ******** * *********** ***** *******, from ******* ** *** ********* *** adjusting ******** ** ******* ****** ** their ********* ***. ** ******** **** process ** *** ***** *****.

Issues **** ******* ***** ********** ** ******

** *** *******, **** ******* **** in *** *****, *** ************** ***** switched ******* "***" ** "******", *********** filling *** ***** **** **** ******** events. *****, **** *** *** ** sitting **** ****** ****, ** ** classified ** * ******, *** **** it ***** *** **** ****** *** camera ******* *** ****, ** ** classified ** * ***.

*********, ** * ****** ****** ** the *****, **** **** ** ********** as * ***.

Poor ****** ********

** ***** **** *****'* ********* ****** tracked ******** ****** *** *****, *********** confidence ** **** **** ** ***** drop *** ******* *** **** **** back **, ******* * *** ****** to ** ******** *** ******** ****** on *** **** ******.

False ********* ** ****** ***** ** *** *****

*****'* ********* ******** ******** ***** ********* in ****** ***** ** *** ***** without ********** **** *** ***** ******* to.

***** ********* ** ****** *** ********* to *** *** ****** **********. **** reduced *** ****** ** ****** ******, though **** **** ***** *******.

False ********* ** ****** ******** ****** ************* **** **** ******** *******

** *** ******** *******, ***** *** issues **** ******** ***** ****** ** parked ******** ** *** *****. **** their ****** ********, ***** ***** ****** were ************* ******* **** ******** ** ~1 ***** ****.

**** **** *** ********* ****** ***** parked ******** **** ********** ****** ** default, ****** ** ***** **** **** further, *** **** **** ****** ******** that ***** *** ********* ** ****** an ***** ** * ******** ******.

PPF ****** *** ****** ********* **** **** *******

** ***** **** ***** ********* ******* ~18-20 *** *** ******** ****** *********, which ** ****** **** *** ******* ~10-12 *** ****** ** ***** ********* in *** *******.

No ***** ****** ** ****

**** ******** **** *** **** ** falling ** ***** ** *** ****** did *** ******* ***** *********.

No ***** ****** ** ******* ******* ** ****** *******

***** ******* ****** ******* ** ***** positives, **** ** ******* ******* ** the ****** ******* ********.

No ****** **** ******* ********** ** *** *****

******* ********** ****** ******* *** ***** did *** ******* *** ***** ********* where **** *********** ******* ****** ******* or **** **** ** *** *****.

People ******** *******, ********, ******** *******

****** ******* ** ******** ******* *** scene **** ******** ******** *** ********** as ******.

People ******** **********

***** ******** ******** ****** **** *** lower **** ** ***** **** *** obstructed **** ******* ****** ** ****** such ** * *** ** ****.

No ***** *** *********** **** ****** ********

**** *** ****** ********, ****** ********* still ** *** ********* **** *** VMS *** ***** ***.

Versions ****

*** ********* ******** **** **** ****** testing.

• ***** **-********: *.*.*.**-******