COGNEX VIDI DEEP TECHNOLOGY
Cognex ViDi deep learning technology allows the automatic inspection of complex
fabrics
ViDi offers the first ready-to-use Deep Learning-based software dedicated to industrial
image analysis. ViDi Suite is a field-tested, optimized and reliable software solution
based on a state-of-the-art set of algorithms in Machine Learning. It allows tackling
otherwise impossible to program inspection & classification challenges. This results in a
powerful, flexible and straightforward solution for countless challenging machine vision
applications.
This novel approach combines the human performance and flexibility in visual
inspection with the reliability and consistency of a computerized system. It finally offers
a way to precisely and repetitively inspect all sorts of natural and man-made textiles.
Cognex ViDi™ radically differs from traditional machine vision solutions since it is:
Self-learning No software development required
Human-like Outperforms the best quality inspectors
Powerful Solves hard to tackle or even impossible to program inspection
challenges
CHALLENGES IN TEXTILE INSPECTION
Complexity
The fabric pattern can be highly complex, and position variants can preclude the use of
simple methods based on spatial frequency analysis.
Natural variability
Visual appearance varies drastically due to the stretchable nature of the fabric and other
variations such as yarn thickness.
Countless forms and types
Defects in textiles come in countless forms and types and explicitly searching for all
defects is not an option.
With ViDi Suite, the automatic inspection of complex pattern fabrics is now extremely
simple. The software algorithm trains itself on a set of known good samples to create its
reference model.
Once this training phase is completed, the inspection is ready to go. Defective areas of
the fabric can quickly be identified and reported. And best of all, there is no need for
extensive defect libraries.
AS SIMPLE AS 1-2-3
ViDi Blue-Locate finds complex features and objects by learning from
annotated images. Self-learning algorithms locate parts, count translucent
glass medical vials on a tray, and perform assembly verification checks on kits
and packages.
ViDi Red-Analyze is used to detect anomalies and aesthetic defects. Be it
weaving, knitting or braiding problems, incomplete or improper stitching or
even printing alignment errors; the red tool can identify all of these and many
more problems simply by learning the normal appearance of an object
including its significant but tolerable variations.
The Red-Analyze tool is also used to segment specific regions such as
defects or other areas of interest. Be it a specific foreign material on a medical
fabric or the cutting zone on lace; the red tool can identify all of these regions
of interest simply by learning the varying appearance of the targeted zone.
ViDi Green-Classify is used to classify an object or a complete scene. Be it
the identification of printed pattern families, the wear-and-tear rates on textile
samples or the separation of acceptable or unacceptable defects; the green
tool learns to separate different classes based on a collection of labeled
images.
ViDi Purple-Read deciphers badly deformed, skewed, and poorly etched
codes using optical character recognition (OCR). The pretrained font library
identifies most text without additional programming or font training for fast,
easy implementation. This robust tool can be retrained to adjust to specific
OCR application requirements—no vision expertise required.
YARN DYE PLAID
For this first woven fabric, we provided our ViDi red tool with a representative set of
good samples to learn the weaving pattern, yarn properties, colors and tolerable
imperfections. After the training phase was completed, the inspection was able to
quickly identify defects like the ones shown.
Weaving weft float
Unexpected stitching
WEAVING
Cognex ViDi
distinguishes
unacceptable defects on
seat belt and tire fabric
while tolerating naturally
occurring variations.
KNITTING
Cognex ViDi identifies a
broken needle in thick
yarn as well as knitting
loops in warps and wefts
despite the complex
pattern
BRAIDING
Cognex ViDi is able to
find various types of
defects such as gaps or
misalignments despite
the reflectivity of the
carbon fibre fabric. The
second example below
shows the detection of a
broken wire on a braided
metal sleeve
FINISHING
Cognex ViDi can detect
anomalies like soil or ink
spots on garments as
well as defects in highly
critical stitching such as
on airbags.
It also excels at verifying
embossed characters on
medical fabrics.
PRINTING
Cognex ViDi allows the
inspection of printed
webs. It is able to identify
problems in printing
quality such as
misalignment of different
color channels (black
outline versus filling)
while the motifs can be
highly complex
TECHNICAL SPECIFICATION
Graphical & application programming interfaces
HTML based graphical user interface (GUI) - Required browser: Mozilla Firefox
Microsoft .Net library
Custom HTTP requests
C runtime library
Hardware & OS Requirements
Dedicated PC for ViDi Suite
CPU: Intel core i5/i7/Xeon NVidia Graphic Card (GeForce GTX770-780Ti-970-980, GTX
TITAN, Quadro K5000-K6000, Tesla K20)
8GB Memory
100GB Disk Space
1 USB port (2.0 or above)
OS: Windows 7 - 64
Support & Maintenance
All licenses are permanent and do not require maintenance or renewable fees
ViDi Suite comes with a free 12 months update & remote application engineering
support
Miscellaneous
License dongle (USB 2.0 or above) included
Supported languages: EN/FR/DE
ViDi Suite Help: EN
Note: ViDi Suite performance - in term of processing time - will depend upon hardware
selection
