COGNEX DEEP LEARNING

Easy-to-Deploy Deep Learning-based Solutions for the Automotive Industry

THE GLOBAL LEADER
IN MACHINE VISION AND INDUSTRIAL BARCODE READING

$1.04 OVER 41
Cognex,® the leading supplier of machine vision
and industrial barcode reading solutions.
With over 3.5 million systems installed in facilities YEARS IN THE BUSINESS
around the world and over forty one years of

BILLION 500+
experience, Cognex is focused on industrial machine
vision and image-based barcode reading technology.
Deployed by the world’s top manufacturers, suppliers
and machine builders, Cognex products ensure
that manufactured items meet the stringent quality
2021 REVENUE CHANNEL PARTNERS
requirements of each industry.
Cognex solutions help customers improve GLOBAL OFFICES IN
manufacturing quality and performance by eliminating
defects, verifying assembly and tracking information
at every stage of the production process. Smarter
20+ COUNTRIES
3,500,000+
automation using Cognex vision and barcode reading
systems means fewer production errors, which equates
to lower manufacturing costs and higher customer
satisfaction. With the widest range of solutions and
largest network of global vision experts, Cognex is the
best choice to help you Build Your Vision.™
SYSTEMS SHIPPED

2 Visit www.cognex.com/products
 DEEP LEARNING-BASED SOLUTIONS
FOR THE AUTOMOTIVE INDUSTRY
Automate complex applications – no machine vision or Piston Ring Inspection.......................................... 4
programming experience required

Cognex Deep Learning is an image-based analysis Cylinder Inspection.............................................. 5

software designed to automate and scale a wide
range of applications across the automotive Welding Seam Inspection................................... 6
industry. Combining human-like intelligence with the
robustness of machine vision, Cognex Deep Learning Spark Plug Identification and Classification.... 7
performs judgment-based part location, inspection,
classification, and character recognition tasks with Airbag Inspection................................................. 8
high accuracy and repeatability.
Trim Final Assembly Verification....................... 9
Using an intuitive interface and common development
platform, the software simplifies application
Vehicle Identification Number (VIN)
development, allowing even non-vision experts to
Code Inspection................................................... 10
deploy reliable automation. Cognex Deep Learning can
easily be adapted to meet changing needs, providing a
Cognex Deep Learning........................................ 11
future-proof solution that grows with your business.

Cognex Deep Learning Solutions for the Automotive Industry 3

 PISTON RING INSPECTION
The defect detection tool simplifies the automated detection and
characterization of defects on textured metal surfaces

CHALLENGE
A piston’s compression rings serve several functions in a reciprocating engine, sealing the combustion chamber and
regulating oil consumption. Defects on a compression ring are difficult to detect because of the piston’s reflective
metallic surface. The cylindrical shape of the piston sometimes renders as blurry and un-focused in images.
Normal variations in the metal’s surface texture are to be expected as part of the manufacturing process, and
some—including rust spots, white areas, and even surface cracks and fissures--are permitted to pass inspection.
Long scratches that affect the piston’s performance and threaten compression levels in the cylinder, however,
are indications of true defects. An inspection system must be able to tolerate normal variations and insignificant
anomalies on compression rings’ surfaces while identifying any long scratches.

SOLUTION
Programming an inspection of this complexity into a rule-based algorithm would require complex defect libraries.
Human inspection, though more flexible, would be too slow. Cognex Deep Learning offers an effective inspection
solution, combining a human’s ability to appreciate minor variations with the reliability, consistency, and speed of an
automated system.

Using the defect detection tool in supervised mode, an engineer trains the deep learning-based software on a
representative set of known “good” and “bad” compression ring images. A technician annotates known “bad” images
where long scratches occur and “good” images with normal variations and tolerable defects, including rust spots and
small cracks. Based on these images, the application learns a piston’s natural form and surface texture, as well as the
normal appearance of scratches. Additional images can be added to the training set during validation testing to reflect
additional examples and help optimize the system. Parameters can continually be adjusted during the training phase
and validation period until the trained model correctly detects and segments all images with long scratches.

During run-time, the software characterizes images with long scratches as defective, having learned to recognize
and ignore irrelevant variations.

Cognex Deep Learning identifies defects on a piston’s compression ring.

4 Cognex Deep Learning Solutions for the Automotive Industry

 CYLINDER INSPECTION
Defect detection tool reliably detects pores in metal

CHALLENGE
The cylinder block is the foundation of an automotive engine. Its large cylinders are the primary working parts of a
reciprocating engine, built to hold the pistons as they pump up and down under compression. Cylinders are normally
made of cast metal and are sometimes lined, or “sleeved,” in a lubricating coating. The cylinder walls must be
durable as they make contact with a piston’s compression rings.

The cylinder can tolerate small irregularities in its metals, which are smoothed out during mechanical wear, but must
not contain bubbles in the metal known as “pores.” These defects are especially challenging to inspect because a
cylinder’s surface is rough, and images appear blurry around the edges due to the depth of field. Specular lighting,
or glare, on the cylinder’s reflective surface also complicates the inspection. It is difficult to program an automated
inspection that tolerates so many small variations in feature shape and location, as well as glare and blurriness.

SOLUTION
Cognex Deep Learning quickly identifies pores when other methods struggle to inspect under the same lighting
conditions. Within minutes, an engineer can train the software on a representative set of “good” and “bad” images
of a cylinder, adjusting the region of interest with a masking filter to eliminate the bright disk of negative space in the
shaft. Using the defect detection tool in supervised mode, a technician annotates the pores in the images labeled
as “bad” and adjusts parameters, including feature size, scale, aspect ratio, and shear to help the model account
for variations in appearance. “Good” images which depict normal cylinders help the software learn which types of
minor casting anomalies and variations are tolerable. The engineer can retrain the system, adjusting parameters
and adding additional images, until the model can generalize the normal appearance of a cylinder and recognize
the abnormalities. During run-time, the deep learning-based software inspects each image within milliseconds,
characterizing those with pores as defective and the rest as normal.

Cognex Deep Learning distinguishes challenging defects on reflective cylinder blocks.

Cognex Deep Learning Solutions for the Automotive Industry 5

 WELDING SEAM INSPECTION
Defect detection and classification tools simplify the automated inspection and classification
of welding seam defects

CHALLENGE
Cognex Deep Learning inspects the integrity of critical powertrain components like pistons, whose complex surface
texture make traditional machine vision inspections difficult. A piston’s welding seams are highly variable, making
abnormalities difficult to distinguish. Certain welding anomalies, like missing, overpowered, or underpowered welds,
are unwanted. Other anomalies, like overlapping seams, are desirable and required for safety reasons. Dark image
areas introduce additional complications. Given the many possible flaws and lighting challenges, deep learning-
based analysis offers a simple and robust alternative to traditional machine vision inspection.

SOLUTION
With Cognex Deep Learning the automated analysis of metal piston welding seams becomes simple. The engineer
trains the software with the defect detection tool in supervised mode on a set of “bad” images representing all
welding anomalies, including overlapping seams, and on “good” samples without any anomalies. In this way, all
anomalies—both those that are desired as well as those that are a cause for rejection—are identified as defects.

In the second part of the inspection, the engineer uses the classification tool to classify seam defects by type. Based
on the model developed during supervision, the software extracts information about specific defects and separates
overlapping seams into their own class. By using these tools together, the automotive manufacturer is assured that
the inspection system identifies all welding seams and successfully classifies the overlapping seams. Based on this
information, the manufacturer is able to select only the overlapping seams for use.

The defect detection tool finds welding anomalies, then the classification tool separates them into useable or unuseable types.

6 Cognex Deep Learning Solutions for the Automotive Industry

 SPARK PLUG IDENTIFICATION
AND CLASSIFICATION
The assembly verification and classification tools identify, count, and classify parts based on their
appearance for kitting/pre-assembly verification

CHALLENGE
For certain identification, counting, and classification applications, manufacturers must rely upon visual inspection
when their environments do not support barcode reading technology. Slight variations in appearance can cause
complications for an automated inspection system. This is the case for spark plugs, which arrive for pre-assembly
on differently colored trays. The inspection system must successfully identify, count, and classify differently colored
spark plugs while ignoring the background color of their trays. This information is then communicated to a vision-
guided robot for assembly.

SOLUTION
Cognex Deep Learning generalizes the distinguishing features of a spark plug based on its size, shape, and surface
features. With the assembly verification tool, an engineer fixtures tray images, teaching the software to identify and
count individual spark plugs. The classification tool uses the deep learning-based model to classify the spark plugs
by the feature germane to the robot—its color.

Cognex Deep Learning finds the position of spark plugs in trays and sorts them by color.

Cognex Deep Learning Solutions for the Automotive Industry 7

 AIRBAG INSPECTION
Defect detection tool inspects textiles for cosmetic defects

CHALLENGE
Airbags are subject to strict quality standards to ensure passenger safety. Automotive manufacturers must double
and triple check all safety-critical components to ensure quality, decrease warranty costs, and reduce recall liability.
This is especially important for airbags, which must be inspected for holes, rips, tears, and seam and stitching issues
that could cause them to fail. These kinds of quality issues are often missed or hard to detect in manual inspection.
They are also difficult to program into a traditional machine vision system because of an airbag’s complex textile
surface. The fabric pattern can be highly complex, and the visual appearance between airbags varies drastically due
to the stretchable nature of the fabric, yarn thickness, and countless small tolerable variations. Because explicitly
searching for all defects is too complicated and time consuming, Cognex Deep Learning offers a simple solution to
identify all anomalous features, without training on “bad” images.

SOLUTION
An engineer uses the defect detection tool in unsupervised mode to train the software on a set of “good” airbag images
to build a reference model of an airbag. The model learns the normal appearance of an airbag’s fabric, including
weaving pattern, fabric properties, and color. All features that deviate from the model’s normal appearance are
characterized as anomalous. In this way, Cognex Deep Learning reliably and consistently detects all anomalies, such
as holes, rips, tears, and unusual stitch patterns. Defective areas of the fabric can quickly be identified and reported
without the need for extensive defect libraries.

Cognex Deep Learning detects all types of manufacturing anomalies in airbag fabric.

8 Cognex Deep Learning Solutions for the Automotive Industry

 TRIM FINAL ASSEMBLY VERIFICATION
The defect detection tool confirms the presence and placement of components on a confusing background

CHALLENGE
The various pieces of trim involved in final assembly verification introduce a high degree of complexity that
challenges traditional machine vision inspections. Human inspectors verify that all parts, such as wire bands
and metal housing, are present and correctly assembled. Subtle lighting contrasts make it difficult to tell whether
the bands are in their correct housing. Human inspectors, though skilled at identifying wire bands, are slow and
inefficient. Cognex Deep Learning uses deep learning-based image analysis to learn the finished appearance of a
piece of trim and identify missing bands as accurately as a human inspector, but with the speed and reliability of an
automated system.

SOLUTION
Using the defect detection tool in supervised mode, a technician trains the system on “bad” images of trim where
the wire is absent, as well as known “good” images where the wire is present, to create a reference model for a
complete piece of trim. Using this model, the application identifies trim pieces with missing wire bands as anomalous
and defective, failing them during final inspection.

Wire present No wire

Cognex Deep Learning identifies missing wire trim in a visually confusing setting.

Cognex Deep Learning Solutions for the Automotive Industry 9

 VEHICLE IDENTIFICATION NUMBER
(VIN) CODE INSPECTION
The OCR tool recognizes deformed characters

CHALLENGE
A vehicle identification number, or VIN code, is a multi-character code used as a unique identifier for automobiles.
VIN codes contain letters and numbers and may be a direct part-marked (DPM); etched or scribed onto a
metal plate; or printed on a sticker. Automotive manufacturers must be able to locate and decode VIN codes for
successful traceability. Specular light, paint colors, and glare make it difficult for a machine vision system to locate
and recognize characters. An inspection system needs to tolerate reflective surfaces which challenge image
formation in order to successfully decode characters.

SOLUTION
With the OCR tool, it is now easy to locate and read deformed characters, despite image formation challenges. To
train the software, an engineer defines the region of interest on images which contain a representative set of VIN code
characters. The tool’s pre-trained omni-font capabilities recognize characters even if they are obscured by glare and
contrast. During training and validation, a technician re-labels only the missed characters until the software’s model
correctly identifies all characters. This new deep learning-based approach to OCR saves time during training and
development by reducing excessive labeling and successfully reads characters on very noisy backgrounds.

Cognex Deep Learning recognizes deformed VIN code characters despite image formation issues.

10 Cognex Deep Learning Solutions for the Automotive Industry

 COGNEX DEEP LEARNING SOLUTIONS
Cognex Deep Learning optimizes productivity by leveraging state-of-the-art technology based on field-tested,
industry-proven machine learning algorithms. Rather than following a rule-based approach to solving inspection
challenges, like traditional machine vision, Cognex’s deep learning solutions learn to spot patterns and anomalies
from reference images. Deep learning automates and scales complex inspection applications that, until now, still
required human inspectors such as defect detection and final assembly verification.

In-Sight ViDi
Simplify complex applications

In-Sight® ViDi™ deep learning tools are deployed on the

In-Sight D900 smart camera without the need for a PC.
The tools are trained using the familiar and easy-to-use
In-Sight software platform, which simplifies application
development, expedites factory integration, and delivers
the power of deep learning to non-vision experts.

ViDi EL
Deploy reliable automation in minutes

Using a pre-trained set of powerful deep learning

algorithms, ViDi EL tools automate complex tasks, like
advanced segmentation and multi-class classification,
quickly and reliably. The tools can be trained in
minutes, using a few as five to ten images, with no
coding required. This, plus an intuitive graphical user
interface and a fast, easy workflow, make ViDi EL tools
accessible to a range of skill levels.

VisionPro Deep Learning

Easily solve tasks — no programming needed

For more challenging inspections, VisionPro Deep

Learning software combines a comprehensive machine
vision tool library with advanced deep learning tools
inside a common deployment framework. It streamlines
development of highly variable vision applications and
allows engineers to build flexible, highly customized
solutions tailored to their specific requirements, all
without the need for complex programming.

Cognex Deep Learning Solutions for the Automotive Industry 11

BUILD YOUR VISION
2D VISION SYSTEMS
Cognex machine vision systems are unmatched in their ability to inspect,
identify and guide parts. They are easy to deploy and provide reliable,
repeatable performance for common to complex tasks.
www.cognex.com/machine-vision

3D VISION SYSTEMS
Cognex laser profilers and area scan 3D vision systems provide ultimate
ease of use, power and flexibility to achieve reliable and accurate
measurement results for the most challenging 3D applications.
www.cognex.com/3D-vision-systems

VISION SOFTWARE
Cognex vision software provides industry leading vision technologies,
from traditional machine vision to deep learning-based image analysis,
to meet any development needs.
www.cognex.com/vision-software

BARCODE READERS
Cognex industrial barcode readers and mobile terminals with patented
algorithms provide the highest read rates for 1D, 2D and DPM codes
regardless of the barcode symbology, size, quality, printing method or surface.
www.cognex.com/barcodereaders

Companies around the world rely on Cognex vision and barcode reading
solutions to optimize quality, drive down costs and control traceability.

Corporate Headquarters One Vision Drive Natick, MA 01760 USA

Regional Sales Offices

Americas Hungary +36 800 80291 Asia
North America +1 844-999-2469 Ireland +44 121 29 65 163 China +86 21 6208 1133 © Copyright 2022, Cognex Corporation.
Brazil +55 11 4210 3919 Italy +39 02 3057 8196 India +9120 4014 7840 All information in this document is subject to change
Mexico +800 733 4116 Netherlands +31 207 941 398 Japan +81 3 5977 5400 without notice. All Rights Reserved. Cognex,
Poland +48 717 121 086 Korea +82 2 539 9047 In-Sight, and VisionPro are registered trademarks of
Europe Spain +34 93 299 28 14 Malaysia +6019 916 5532 Cognex Corporation. ViDi is a trademark of Cognex
Austria +49 721 958 8052 Sweden +46 21 14 55 88 Singapore +65 632 55 700 Corporation. All other trademarks are property of their
Belgium +32 289 370 75 Switzerland +41 445 788 877 Taiwan +886 3 578 0060 respective owners. Lit. No. ViDiAuto-04-2022
France +33 1 7654 9318 Turkey +90 216 900 1696 Thailand +66 88 7978924
Germany +49 721 958 8052 United Kingdom +44 121 29 65 163 Vietnam +84 2444 583358 www.cognex.com