How to use Video AI at the Edge Magic Pao, Director, Advantech - - PowerPoint PPT Presentation

how to use video ai at the edge
SMART_READER_LITE
LIVE PREVIEW

How to use Video AI at the Edge Magic Pao, Director, Advantech - - PowerPoint PPT Presentation

How to use Video AI at the Edge Magic Pao, Director, Advantech Industrial IOT Industrial AI Business GAP? Edge Technical SI Domain Focused SI Customers Computing Application Field Installer Stress Testing Domain know-how SDK &


slide-1
SLIDE 1

How to use Video AI at the Edge

Magic Pao, Director, Advantech Industrial IOT

slide-2
SLIDE 2

Industrial AI Business GAP?

AI Box Data Pre- Processing SDK & Training Model Domain know-how Stress Testing Field Installer Application

Edge Computing Technical SI Domain Focused SI Customers

slide-3
SLIDE 3

Pre-built AI Models

(ISV Partners)

Cloud Services

AI Training Computing System AI Edge Inference System

People Detection Vehicle Classification PCB AOI Agriculture Sorting Textile AOI

Code IDE Hyper Parameter Data Pre-Processing Model Management Inference Engine Deploy Compute Resource Mgmt.

AI Framework Service SKY-6100 SKY-6200 SKY-6400 SKY-6420 AIR-100 AIR-200 MIC-710IVA MIC-720AI VEGA-320 VEGA-330 MIC-770 MIC-730AI

Bring AI to the Edge

Edge AI Computing Platform

slide-4
SLIDE 4

What Advantech Role is – Edge AI Computing

Environmental Fits

Harsh Environment Consideration Customized Design

slide-5
SLIDE 5

What Advantech Role is - Edge Computing

  • Remote Update
  • Operating in Equipment
  • Peripherals Support (Lighting, Motion…)
  • Video Capture Protocol
  • Industrial Lifecycle Management

Domain Pain Point

slide-6
SLIDE 6

What Advantech Role is - Exposure

Industry 4.0 & Automation Exhibition Transportation & Smart City Medical & Equipment

Vertical Application Events

slide-7
SLIDE 7

What Advantech Role is - Exposure

Cross-Industry Link

slide-8
SLIDE 8

SKY-6100 1U 5xT4 SKY-6200 2U 4xV100 SKY-6400 4U 4xV100 SKY-642 4U 10xV100

What Edge AI Product

– AI Training Computing(5G, Server)

520 TOPS

  • Low-latency interactions
  • Reduced bandwidth to the cloud
  • Data privacy
slide-9
SLIDE 9

What Edge Product

Edge AI Systems AI NVR

Industrial Healthcare Traffic Monitoring Robot Guidance Manufacturing Factory MIC-730AI MIC-720AI MIC-710AI MIC-730IVA MIC-720IVA Casino Logistics Smart City

– Edge Inference (Industries, NVR)

slide-10
SLIDE 10

What Edge Product--- Industrial AI Dev. Box

QCap Edge Inference System QDeep

  • HDMI/SDI/POE
  • H.265/H.264
  • RTSP/RTMP/HLS
  • Jetson Nano
  • Jetson TX2
  • Jetson Xavier
  • Facial/Object Recognition
  • Counting/Timer
  • Behavior/Gesture

MIC-710AI MIC-720AI MIC-730AI

Capture/Record/Stream Inference/Decode/Monitor Identify/Analysis/Tracing

JETSON NANO JETSON TX2 JETSON XAVIER

5 - 10W 0.5 TFLOPS (FP16) 7 - 15W 1.3 TFLOPS (FP16) 10 - 30W 10 TFLOPS (FP16) | 32 TOPS (INT8)

Video Input DLI APP

slide-11
SLIDE 11

Why Industrial AI Dev. Box? AI Training Course

What they learned is What they can make…

Picture source: Nvidia

slide-12
SLIDE 12

Why Industrial AI Dev. Box? AI App. Developer

“AI APP” Developers……

slide-13
SLIDE 13

Why Industrial AI Dev. Box? AI Dev. Lab

“AI + Automation” Integration…

slide-14
SLIDE 14

Edge AI Deployment Strategy

– Technology Customization

Objective Feasibility Study

  • Performance (Hardware)
  • Accuracy (DLT)

Automation Installation Image Data Capture

Golden Process

AI is Everywhere…

Tech Customization
slide-15
SLIDE 15

Edge AI Deployment Strategy

– Cloud-based Management WISE-PaaS/AFS

Cloud Management
slide-16
SLIDE 16 Solution Ready

Edge AI Deployment Strategy

– Solution-Ready

Edge AI Deployment Strategy

Phase-1: Prepare

  • Webinar
  • Hands-on DLT Training

course

  • POC application

Phase-2: PoC

  • Data Pre-Processing
  • DLT & accuracy report
  • Go or No-Go

Phase-3: DLT

  • Image Capture Setting
  • Modeling, and Training
  • Accuracy Turning
  • Workflow setting, and

architecture setting

Phase-4: DLI

  • Technical transfer
  • Local Installation
  • AFS Auto-Deployment(Optional)
slide-17
SLIDE 17
slide-18
SLIDE 18

Co-Creating the Future

  • f the IoT World

Co-Creating the Future

  • f the IoT World