Deeping Source AI Startup Profile
Deeping Source is a Korean startup led by ex-Intel engineers that is targeting retail video analytics, including complete anonymization while maintaining usability with AI, gaze estimation, and compatibility with existing systems.
In this report, based on an interview with company executives, IPVM examines the company's background, approach, positioning, and challenges.
**********
******* ** **** ****, ******* ****** entered *** ****** ****** **** ******** data ************* ********** *** ***** *********, that *** ****** ********** ************ *********** from *** **** ***** *********** *** utility *** ******* ******** *** ** applications. *** *******'* ******* *** ******-**** ***, **.*.*** *** **-******* ** *** ****** recognition *******, ********, ********** ******** ** **** *** $** million.
*** ** ** ***** ******* ******* *********** *** ** ********* ** ******** for **** **** * ******.
*******
******* ****** **** **** ********* $* million ** ******* *** ****, ****** coming **** ***** **********. ** ********, Deeping ****** *** ****** $** ******* in ** ******* ** ** ******* 2022 *** ******* "**** ** ******** of *******" ** ******* ** ****.
Accelerator *******
******* ****** *** **** ****** ****** ** *** Fuel *********** *******, ****** ** *** ****** ********** and ****** ** *********** ********.**** ************* ******** ** **** ** **** not **** ****** ** ************ ***** focusing **** ** "******** ********** *****" including ************* ** ****** **** ********* like *******.
*********
*** ******* ********* ********* ******, **** ** ******* ** South *****, *** * ** *** US. ******* ****** *** ** *********** core, **** **% ** *** ********* working ** ******* ******** *********** ** software ***********. ******* *** ****** ** funding, *** ******* ** ***** ********** small.
*******
******* ****** ************ ********* **** ******** *************.*** ******* ******** *****-****** *****-****** ********, gaze **********, *** ********, ** ********* from ********** ****, *** ****** ****-****** dashboards.
*** ********* ******* ****** ************ ***** details ***** *******:
******* ******** ****-*** (*****) ************ *** *******'* *********.
*** *** *************** ** *** ******* is **** ** ******** *** **** anonymization **********, ***** ***** **** ** remove *** *** ******** *********** ** the ***** ** ** *** ***** while ********** *** ******* ** *** AI. ** *** ** *** ******* differences ** *** ********** ** ****, even ****** ** ****** *** *** visual *** ******** ***********, ******** **** some *********** *** *** **, ** can ***** *** *** ** **** a ****** ********* *** ******** ********* and ******** ******. ** ** ** use *** ******** ***** ******.
******* ****** ********* ***** **** "*** existing ****** ** **** ** ******** because ** **** *** ****** ************* solutions ** ** *** ****** **** using *** ******** **** *** **** a ******** **** ** ****." *** company ****** ********* ********* ** **** on **** ** ******* **** *** integrated **** *********' *********** *******.
** ******** ** **** *************, ******* Source ****** *****-****** ******** ***** ******** cameras *** **** **********. *** *******'* GM ** ***** *******,***** ********, ******** *** *** ******** **** their *********.
*****'* * ****** ** **** ******, video *************, *** **** ******** **** we *** ** ** **** **** retailers ********** **** *** *** ***** people *** ******* ** * *****, for *******, ** * *** **** of **. *****'* **** ***** **** I ***** * *** ** ****** do **** ** *** **** ********** from ****** ***** **** ********* ** a ******** *********, ***** **********, ***** analysis *** *********** *********. *** ***** that ** ********** ******* ****** **** the ****, ** ***** ******; ***, the **** ************* ** **** ***** stream ******* **'* *** * ***** mask, ***, *** **** ********, *** three, *** ******* ** **** ********* of ***** **** ** *** ******** camera ********* ****** *** ******* *** reuse ** **** *** **** *** caveat. ** **** ** ** **** to *** * ****** ****** *** a ********* **** ****** **** ** our *** ******* ***** *** ******* than *** ****** ******* ** **** less ********** *****.
*** ********* ****** *********** *****-****** ********:
** **** ** **** ********** **** by ******* ******:
******* ******'* ** ******** ******* ********* their ******** ** *** ******.
** **** ****** ******* ** *** retail ******. ** *** **** ******* at *******, ********, ****** ********, ***** commercial *********, *** ****** ****** ******. It's **** **** **** ******.
Opportunity *** ***********
*** **** *********** ** ********* ******** analytics *** ********** ************ ***** *** main ********** ** *** ********* ** purchasing **** ********** *** ** ******** to ******** ************ **************.
***** ** **** *** ****** ***** analytics, ** ****** ***** ** ***** performance. ************, ******, ******* ****** *** an *********** ** ************* **** ********* such ** **** ********** *** *****-****** tracking **** ** *** ******** ******* with ************ ******* *** *********.
****, *** *********** ** **** ******* Source's ************* ********* *** ******** ** others (*** **** ** ****** **** the **** *** ***** *** *** work ****, ******, ***** ** **** not ****** ******* ******'*). *** ***** issue **** *** ************* ********* ** how ********* **** **** **. ***** it ***** ** ***** ********* ** the **, ********** ** **** *******-********* markets **** ** *******, ** *** skeptical **** ***** ******* ** *** enterprises **** **** **** ** ********.
******* ****** **** ** ******* ** enterprises ****** ** ******* ************* *** likely ** **** *** ***** **** and ******* ***** ** ******** ********* from ********* ******* **** ******** ** Hanwha ** *******, ***., ** ** "good ******" ** **** *** *********** free.
*******
**** *** ******* ********* ***** ** North ******* *** ******, ******* ****** is ********** ** **** *** ******** in *** **** *** *****, ***** targeting *** ******* **** **** **** anonymization. *******, ******* ******'* ******* ********* target * ***** ******, *** **** significant *********** **** ***** ********* ********* with ***** *********, *** ***** ******* Source *** ****** ** ***** ******** and ******** ******* ******* ** ** seen.