Machine Vision & AI Realizes Automated Inspection in - - PowerPoint PPT Presentation

machine vision ai realizes automated inspection in
SMART_READER_LITE
LIVE PREVIEW

Machine Vision & AI Realizes Automated Inspection in - - PowerPoint PPT Presentation

Machine Vision & AI Realizes Automated Inspection in Manufacturing Moderator: Ethan Huang, IIoT Advnatech Panelist 1: Yuki Tominaga, Phoxter (Japan) Panelist 2: Neil Chen, IIoT Advantech Panelist 3: Jesse Chen, CEO SmaSoft (Taiwan) Dec,


slide-1
SLIDE 1

Dec, 2019 IIoT Advantech

Machine Vision & AI Realizes Automated Inspection in Manufacturing

Moderator: Ethan Huang, IIoT Advnatech Panelist 1: Yuki Tominaga, Phoxter (Japan) Panelist 2: Neil Chen, IIoT Advantech Panelist 3: Jesse Chen, CEO SmaSoft (Taiwan)

slide-2
SLIDE 2

Edge SRP & Go to Market Milestones

Co-Creation

共创

Co-Creation

共创

Machine Vision Solution AI Vision Inspection Solution

2018 2H

2019 2Q 2019 4Q 2019 3Q 2019 4Q 2020

CCJV

slide-3
SLIDE 3

Automotive 19% Packaging 17% Semicon 15% Electronics 8% Pharmaceutical 6% Glass 5% Metal 5% Wood 4% Other Manufactuing 10% Non Manufaturing 11% US$1.6bn

US$2bn

US$160mn US$400mn US$120mn

CAGR: 7.8% 2016-2020

US$680mn US$600mn US$800mn US$440mn US$320mn

Machine Vision Markets & Products

slide-4
SLIDE 4

Processing

1.

Alignment; inspection

Assembly

2.

Guidance; Gauge, Verification

Inspection & Testing

3.

Identification; Defect Inspection

Packaging

4.

Identification; Guidance

Logistics

5.

Identification; Guidance

Future Manufacturing Applications:

1) Structured data extraction, Gauge, Coordinated Pattern Recognition => Machine Vision Solution 2) Randomized image Recognition => AI Inspection Solution

Machine Vision applications in manufacture

slide-5
SLIDE 5

Advantech VisionNavi

  • 1. Market demands for the Software
  • 2. The Solution
  • 3. Application
  • 4. Co-Creation

Agenda

Dec, 2019 Yuki.Tominaga, Phoxter

slide-6
SLIDE 6

Growing needs of Machine Vision

Health, Labour and Welfare Ministry Japan https://www.mhlw.go.jp/english/wp/wp-hw3/dl/j1_05.pdf Report from Tsukubabank https://www.tsukubabank.co.jp/corporate/info/monthlyreport /pdf/2016/03/201603_11.pdf

 Increase of wage  Labor shortage, Education cost

slide-7
SLIDE 7

 Make more money/Save money

– speed up the production – reduce the number of inspectors in the production line – improve quality; by categorizing defects and feedback the result to production

 Protect their own brand

– prevent outflow of defective products to the market – execute reliable inspection

What the users really want to do with Machine Vision?

slide-8
SLIDE 8

VisionNavi

slide-9
SLIDE 9

Industries

Beverages Machine parts Medical Semiconductor Food Electric parts PWB assembly Logistics

slide-10
SLIDE 10

Applications

Measurement Find defects/flaw/spot/etc. OCR Code reading, 1D/2D

  • Preprocessing
  • Filter
  • Color extraction
  • Color convertion
  • Image rotate
  • etc.
  • Judgement
  • Positioning
  • Count
  • Compare with good sample
  • Blob analysis
  • Matching
  • etc.

More features for various situation

slide-11
SLIDE 11

AOI solutions! People with inspection problems Phoxter has solutions for AOI, but no sales link to the people who need them

slide-12
SLIDE 12

Co-Creating the Future

  • f the IoT World

Co-Creating the Future

  • f the IoT World
slide-13
SLIDE 13

Dec, 2019 Neil.Chen, IIoT Machine Vision, Advantech

Go To Market Strategy

slide-14
SLIDE 14

Electronics Manufacturing Process Major MV Application Product

Vision-Based Track and Trace Solutions $3.93 billion, CAGR 18.9% by 2023

PCB/SMT/DIP Assembly Quality Assurance Packaging Logistics Inspection Identification Inspection (High Growth) Device Platform Device Platform Vision Sensor Vision System, Vision Sensor, ID Reader

Source: AIA Online Marketing, 2019 Reference: Bernstein analysis, 2017

Guidance (High Growth)

slide-15
SLIDE 15

Industry Stationary Retail/Industry Handheld Retail Handheld

RFID Vision Laser

Reuse Expensive tag Dimension Stable Low price Reliable High Price Multi-code, OCR

Solutions in Identification

slide-16
SLIDE 16

Edge Computing

Cloud Service Optics Partner Field Device

  • Marketplace
  • OEE Monitoring
  • Visualization
  • Application Driven
  • Domain Focus
  • DFSI
  • Sensor, I/O, Data
  • Control, Comm.

HMI, SCADA

Go To Market

1D Code 2D Code O.C.R O.C.V

Wireless Sensor Module Gateways Vision sensor I/O & DAQ Devices Automation PCs Intelligent HMI Compact IPC

VisionNavi

Value Claim/ Ecosystem Whole Process Material/Finish Good/Assert

slide-17
SLIDE 17

DFSI Optics, Use Case Online Evaluation Evaluation Kit Virtual Machine Trial Test

Integrate optics(Lens, Lighting), duplicate use case Upload images, feasibility study On site demo, verify result

Service Network

slide-18
SLIDE 18

Advantech Confidential & Internal Use Only

Identification: 1D/2D code Gauge/Alignment Present/Counting Segmentation Classification Verification

Rule Base Deep Learning

O.C.R/Defect Detection

Machine Vision Algorithm Mix

slide-19
SLIDE 19

Co-Creating the Future

  • f the IoT World

Co-Creating the Future

  • f the IoT World
slide-20
SLIDE 20

Advantech AINavi

Dec, 2019 Jesse.Chen, SmaSoft

1.Overview, Architecture 2.Selling point

  • 3. Features
  • 4. Applications

Agenda

slide-21
SLIDE 21

Company Introduction 偲倢科技

slide-22
SLIDE 22

OUR COMPANY

SMASOFT

With the emergence of industrial 4.0 brings a tidal wave of smart manufacturing and with this, in contrast to past production line focusing on hardware performances, the current age of smart manufacturing focuses more on hardware software system integration. It is no coincidence that Smasoft’s expertise and focus has been the system integration, both hardware and software, of an automation

  • solution. This is done with the software LabVIEW from

National Instruments (NI), and Smasoft’s proprietary SmaSEQ, that allows users to easily integrate robotic arm controls, automated optical inspection (AOI) algorithms using industrial cameras, and even AI model inferencing all without any prior programming experience.

slide-23
SLIDE 23

OUR WORK

Solve your automation problems

Developing an automated manufacturing system requires highly- trained professional software engineers who, often need a significant amount of time and effort to train; Hence, with a software integration platform like SmaSEQ, one would be able to automate any production line applications without any programming experience, which would surely increase the working productivity and efficiency, and significantly shorten the training/educating time-span required for an automation software engineer.

We are the Automation solution supplier

slide-24
SLIDE 24

OUR SOFTWARE

SmaSEQ

Unlike the traditional production lines in which different hardware have distinct software and programming languages,

  • ur

product

  • ffers

complete design and compilation for the entire industrial automation process as a total solution, which offers many advantages such as ease-of- use when compared with solutions made up of

  • ther

independent equipment; This greatly speeds up hardware integration and software development process, and also greatly enhances production performance.

slide-25
SLIDE 25

Automation Modules Multi- Threads Place notes

  • n scheduler

Scheduling Schedule Functions Easy-To-Use setting

OUR SOFTWARE

“SmaSEQ”

slide-26
SLIDE 26

SmaAITrainer

Standalone AI Model Trainer

slide-27
SLIDE 27

SmaAI

AI Defect Inspection

High-speed AI inference A high-speed and high-accuracy AI solution Distr trib ibuted AI AI stru ructu ture re Simultaneously run multiple AI models cost-effectively

slide-28
SLIDE 28

SmaAI Trainer Training Flow

Data Collection

NVIDIA T4 Advantech

Model Deployment

SmaAI Trainer Training IPC

MIC-770+ MIC-75G20

slide-29
SLIDE 29

Distributed SmaAI Inference System

Vision Motion Robot DIO

DeepStream

… … … … … …

GPU IPC w TensorRT

Model 1

GPU IPC w TensorRT

Model 1

GPU IPC w TensorRT

Model 1

MIC-730AI

slide-30
SLIDE 30

Daily Check

NG OK

Labeled data Upload Data & Re-training Task

4

Marketplace

Public Cloud

Smasoft SRP

Camera / Ruled-base AOI MIC-770

Subscribe

1

Model Update

5

Edge AI System MIC-730AI

Model Deploy

2

OK image

Operate/QA

Result

3rd

rd part

party Pass assive Com

  • mponents Defect Ins

nspectio ion wi with WISE SE-PaaS/AFS

Detectable defect:

Dark crack Missing angle Poor soldering Missing material dislocation

slide-31
SLIDE 31

OUR PARTNER

Smasoft software platforms have been approved by various partners; and with partners’ support, we can supply more innovative and diverse solutions to our customers.

slide-32
SLIDE 32

S T E P 0 1

I m a g e A c q u i s i t i o n

S T E P 0 2 C l a s s D e f i n i t i o n S T E P 0 3

A I T r a i n i n g

  • Image Augmentation and processing
  • Image Data Verification
  • Preview of the results

S m a A I T r a i n e r S m a A I

Edge SRP

slide-33
SLIDE 33

T r a i n i n g C o u r s e S m a A I T r a i n e r

Training Services

SmaAcQ SmaAITrainer

S m a A I

SmaAI Knowledge Base Example Project Classroom Training Online Training WISE-PaaS/AFS

slide-34
SLIDE 34

M a r k e t i n g S u m m a r y L i v e D e m o s

Go to Markets

T r a i n i n g M a t e r i a l s W e b i n a r

slide-35
SLIDE 35

Edge SRP & Go to Market Milestones

Co-Creation

共创

Co-Creation

共创

2018 2H

2019 2Q 2019 3Q 2019 4Q

Welcome to join us

2020 CCJV

slide-36
SLIDE 36

Co-Creating the Future

  • f the IoT World

Co-Creating the Future

  • f the IoT World