Intel Neural Compute Stick 2 / Movidius AI Test

Author: IPVM Team, Published on Nov 21, 2018

AI is a major trend in video surveillance with manufacturers paying significant attention to Intel's Movidius Myriad chips. Indeed, Avigilon has announced that their next generation AI H5 cameras will use the Myriad X.

But how well do these Myriad X chips perform?

IPVM has done its first testing of the Myriad X chip (via the NCS2) and shares results inside this report.

Intel / Movidius / Network Compute Stick Overview

Cameras like Avigilon will use Intel / Movidius chips inside them to deep learning / Artificial intelligence. To the side is an image of a Myriad X chip. Video will be sent into to on-board chip, the chip will run deep learning models and it will output categorizations of people, objects, faces, etc.

Just a week ago, Intel announced its second generation of 'Neural Compute Sticks' (NCS). While the first NCS had the original / older Movidius Myriad 2 chip, this new one has the newer Myriad X inside, which Intel claims "delivers 10X the performance compared to previous generations".

IPVM bought the NCS2 and tested it.

Areas Analyzed

Inside this report, we share results across:

  • Face detection using Intel's face-detection-retail-0004 model
  • Testing across Intel i7, Myriad 2 and Myriad X
  • FPS, CPU and RAM consumption variances across devices
  • Accuracy comparison of detection
  • YOLO support and issues

** ** * ***** ***** ** ***** ************ **** ************* paying *********** ********* *******'* ******** ****** *****. ******,******** *** ********* **** ***** **** ********** ** ** *********** *** *** ****** *.

*** *** **** ** ***** ****** * ***** *******?

**** *** **** *** ***** ******* ** *** ****** * chip (*** *** ****) *** ****** ******* ****** **** ******.

Intel / ******** / ******* ******* ***** ********

******* **** ******** **** *** ***** / ******** ***** ****** them ** **** ******** / ********** ************. ** *** **** is ** ***** ** * ****** * ****. ***** **** be **** **** ** **-***** ****, *** **** **** *** deep ******** ****** *** ** **** ****** *************** ** ******, objects, *****, ***.

**** * **** ***,***** ************ ****** ********** ** '****** ******* ******' (***). ***** *** ***** *** *** *** ******** / older ******** ****** * ****, **** *** *** *** *** newer****** *******, ***** ***** ****** "******** *** *** *********** ******** ** previous ***********".

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

Areas ********

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

  • **** ********* ***** *****'* ****-*********-******-**** *****
  • ******* ****** ***** **, ****** * *** ****** *
  • ***, *** *** *** *********** ********* ****** *******
  • ******** ********** ** *********
  • **** ******* *** ******

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

Key ********

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

  • Myriad * ~= *.** ****** * ***: The NCS2 ran roughly 2.7x frames-per-second more than the NCS1 in their face detection model we tested. Intel says that different optimizations and measurement changes would increase this significantly (discussed in caveats below).
  • Faster **** *** (**): The NCS2 / Myriad X outperformed the i7-7700K (a top-of-the-line in 2017) chip tested here, a significant advance given the Myriad X being significantly lower cost and lower power consumption.
  • Little/no ******** ****: Our tests noted no differences in output video accuracy in line's feedback that OpenVINO's model-optimization step should be considered "lossless." They pointed out that allowing lossy model compression (when converting a model to be run on Myriad; eg FP32->FP16 data-type downsizing) would cause issues with vendors training a model with certain expectations, and having a different (lossy) model for inferring.
  • Model ************* ** ******: While Intel has mentioned YOLO support for OpenVino, the current YOLOv3 is not yet compatible on Myriad chips due to a particular network layer. ***** **************** ***** *** ******* ********** ****** ** ** *** ** CPU, *** ****** ** *** *** ********** ** ****** ******* due ** * ********** ******* *****.
    **** ** & **: ***** ******* ***** *** ** *** on ******, *** *** ***** *******.
    **********+***: *** ***** ** **** ****** **** *******, ************ ***** face-detection-retail-0004 (********** ***** **** * ******-**** ********).
    *********+***: *** ***** *******. ** **** ******* ****** ****** **** Github ** *****'* *** ******.

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

intel performance comparison2

Caveat ** *** / *****

***********, *****'* *** ******* *** **** ****** **** *** ***, for **** *******. **, *****'* ****-*********-******-**** ***** ******* *** *** on ***, **.* ** ******* (**** ***** ******** ********** ******* numbers ** **** ******).

* *** ****** ** *** ***** ***** ******* ***: (*) We *** ******, **** ******* **** ***** *++ ****; (*) our ***-******* ******** ******* **** (** ****** *******/******* ******), *** isolating ***** ****; (*) ** *** *** ***** ******* ************ handling ** ***** ***** *******, ***** ** ***** **** ********.

Video **********

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

Questions / ******** / **** ** **** ****?

** **** ******* ******* ** *** **** / ****** * along **** ***** ********* *** **** **** ****** **** *** next *** *****. **** ***** *** **** ** *** ******?

Comments (12)

*'* **** ** ********** **** **** ** * *** ****. Do *** "****" **** ***** **** * ******** *** *** a ******* *** *** ****** ** ***** *********** *******? *** these ******** **** ****** ** ******** *** *********?

*** **** ****** ** **** * *** **** **** *** other *** ******. ** * **********, ***** ****** ******* ****** aren't ****** *** ********** ************, *** *** **********/******* **** ****** network / ******* ** *** ****** **** (** *** *** it ********, ****** *** **********, ***). ***'** ** ***** ** integrated ****** **** ** *** **** *****.

**** **** ** *** *** **** ** *** **** ***** (I'll *** ***** ******* ** ****-****** ******* ** **), *** it's **** **** **** *** ***. ***'** ** ****** ****** cloning ********'* *** ****** *++ *****. *** *** **** ***** (after *** ******* ********) ** ~/*****/*******************/****************/****************/*******. ** ** "*** **** open ****** / ********", ** ******** ***** ** "***" - they **** ** *** *** *******. *** ***'** **** ** read *** ******** ******* (**** ** *******) ** ~/*****/*******************/*********.

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

******* ********, *** ** ******* $ ** $$ ** $$$?

**** *** $***. ********** **********, *** ****'** *** ****-***** *********

****** **** **** **** **** **-*****. * ******* *** * days *** *** ** ** ******** ******.

**** ******** - **** ***% *** *********** ** *** **-***** mean **** **** ********* *** ***** *** ***** **** **** of ** ** ** ** ******** ******* **** ***** ******?

****, *** *** *** *** **** *** ****** *******? ** there * *********** ****** ** ******* ** *** *****? **** wondering **** *** ****** **** ** ** *** ******* *******/****.

*** **** * ** ***, *** *'* ************* ******. * think **'* ** *** ******, ** ********* ***% **** *********** of * **** *** ** **'* * ***** (** ********** aggregate ****** ****). ** ************ *'* ******** ** **** *.**/* cores *****.

* ***** *** ***** *** *** **** **, ** ********** heat. *'** *** ** ***** ** * ****** *** ** gauging. **** ***** ** * * *** *****, *********** ******* of ****** * **** (** ***** ****** **** *****) ******** to **** ****** ****.

* ***** **'* ** *** ******, ** ********* ***% **** utilization ** * **** *** ** **'* * ***** (** comparable ********* ****** ****).

** **** ** ****** *** ****, * ***** ******* **** the ****** **** ** ** ********** ****** ********, *.*. ***%=**%.

****** **** ******** *** ***%, ***** *** * *** ** cores? ** ** **** ***** ******** **** **** *?

* *** **** ******* ******* ** *** ******* *******. ***** it ** ***** *** ***** *** ********* *** ***(******* ****** Power). **’* **** *** **** *** ****. * ** *** certain *** ** ******** ** *** *** ** * ******** chip **** ** ******* ** ** ******* *** ******** ***** to ******* **** **** ** **** ** ********... ** *** not **** ****. * *****’* **** **** **** ** **** wattage ** ******* **** ** **** ****** *****. **** ** how ****** * ******** ** ***** ****** **********:

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

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

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

****** ***** - ** ** *** **** *** is **** ** ** *** *** **** ******** **** *** server **** ********** ****** ********* *********

* *****’* **** **** ** **** *** *******/******* **** *** mobile ********** **** ********** ** *****. ***** ***** ** ******* everything **** ***. * ******* **** *** *** * **** but ******** **** ******* ***** ** ******* ***********.

***, ** *** ***** ** ** *** / ************ ******** on ** / **** ******** ** *** **** ****** ** well.

** *** **** ******* ** ** ********* ***, ********** **, some *** *** **** **** ***** / ******** ********* ****. Upcoming ********* **** ******* ****.

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

* ****** * **** ** ***** ** **** **** **** mouser ** *** ****** ****** ************ *** *** **** ** play **** **** **** *** *******. ** ******* * *** openVino **** *** ***** *** *** ** **** *** ********* the ***** ** ***(*** ******). ******* ** ********** ** *** it ****** *** *** ** **** ** ******.

(** *** ** * *** ****, *** "***" ****** ** only *** ***** ***'* *** ******, ***.)

* ** ************* ** *++, **** ******* *******, *** *** samples **** ****** ****.

* **** **** ******** *** ******* *** ******* **** ** the ******* **** ** *** **** *********** ** ****. ** is **** ** ********** *** *** *** ****** **** **** other *** *** ****** **** *** ** ***********. *** *** license ***** ***** *** ****** ** ********* **** ** *** into *** ******* ***** ********* ***** ** *** **** *** license ***** ***********. **** **** * *********** ******* ********* *** recognition.

** ***, ****** ************ **** ***********, *** ** ** ****** early ** **** ***** ***.

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

Solink Raises $12 Million - Company Profile on Feb 12, 2019
Most industry professionals have never heard of Solink, a company whose tagline is: It's time to revolutionize the way business uses...
ONVIF Video Surveillance Tutorial on Jan 29, 2019
ONVIF is well known within the surveillance industry as an interface to connect IP cameras and VMS systems. However, new users may find it...
Bosch Favorability Results 2019 on Jan 23, 2019
Bosch's favorability moderately strengthed, in new IPVM integrator statistics over their results from 2017, with 2019 results showing strong net...
Avigilon Favorability Results 2019 on Jan 15, 2019
Since IPVM's 2017 Avigilon favorability results, the company was acquired by Motorola and has shifted from being an aggressive startup to a more...
Gorilla Technology AI Provider, Raises $15 Million, Profiled on Jan 15, 2019
Gorilla Technology is a Taiwanese video analytics manufacturer that recently announced a $15 million investment from SBI Group, saying this...
CES 2019 Show Report on Jan 10, 2019
IPVM was at CES 2019 for the second year (see our CES 2018 Show Final Report) and is reporting on announcements and interesting news from the show...
Managed Video Services UL 827B Examined on Jan 09, 2019
Historically, UL listings for central stations have been important, with UL 827 having widespread support. However, few central stations have...
IPVM Best New Products 2019 Opened - 70+ Entrants on Jan 07, 2019
The inaugural IPVM Best New Product Awards has been opened - the industry's first and only program where the awards are not pay-to-play and the...
Axis Tailgate Detection Tested on Jan 04, 2019
Axis is aiming to tackle tailgating, one of access control's biggest issues, with the Tailgating Detector ACAP application. This camera app claims...
The Battle For The VSaaS Market Begins 2019 - Alarm.com, Arcules, Eagle Eye, OpenEye, Qumulex, Verkada, More on Jan 02, 2019
2019 will be the year that VSaaS finally becomes a real factor for professional video surveillance. While Video Surveillance as a Service (VSaaS)...

Most Recent Industry Reports

Exacq Raises VMS Software Pricing Twice in Less Than a Year on Feb 18, 2019
Most VMSes regularly release new features, but rarely increase their prices. For the 3rd time in 4 years, and 2nd time in 8 months, since being...
Axis IR Multi Imager Camera Tested (P3717-PLE) on Feb 18, 2019
Axis has released their first IR multi imager, the P3717-PLE, a repositionable model listing 360° IR illumination and flexible positioning,...
Ubiquiti Favorability Results 2019 on Feb 18, 2019
Ubiquiti has quietly grown into a $1+ billion annual revenue company, with offerings across wireless, wireline network and video surveillance (see...
Casino Surveillance Pro Interview: James Lathrop on Feb 15, 2019
James Lathrop has been working in casinos for almost 25 years. During that time, he says he has held "just about every job you can do in the...
Hikvision 2018 Revenue Tops $7 Billion USD But Growth Slows To Low on Feb 15, 2019
Hikvision's annual revenue topped $7 billion for the first time in 2018, although growth slowed sharply. In this post, we analyze the latest...
Hanwha Smaller Multi Imager Tested (PNM-9000VQ) on Feb 14, 2019
Hanwha's first repositionable multi imager PNM-9081VQ tested well, but was huge, over 12" wide and weighing in at over 10 pounds. Now, they have...
ADT And 'The Defenders' Silent About Massive Complaints on Feb 14, 2019
ADT's largest dealer, "The Defenders" has been the subject of a massive number of complaints over many years and many forums, most recently a CBS...
Hikvision Chairman Praises United Front on Feb 14, 2019
Hikvision’s controlling shareholder held a meeting last month praising the United Front, a Communist Party organization known for its secretive...
Sales Turnover At Anyvision on Feb 13, 2019
Anyvision raised a $43 million Series A and according to their newest investor: what you need to do is push the gas pedal and build an...
Cisco Meraki Cloud VMS/Cameras Tested on Feb 13, 2019
Cisco Meraki says their cameras "bring Meraki magic to the enterprise video security world". According to Meraki, their magic is their management...

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