Transcriptions
Note: this content has been automatically generated.
00:00:02
good afternoon it's a pleasure to be here and to present you with the
00:00:06
seasons my name is actually recovery time uh replacing and go here
00:00:15
images and videos or more and more important in our professional lives
00:00:20
and you did you have developed a unique keep learning based image analysis tool box which provides you an easy
00:00:26
flexible yet powerful way to extract important information
00:00:31
from these type of images and movies
00:00:34
beat forty us that inspection of the credit watch parts
00:00:39
the track road users on a busy street summer in china
00:00:45
to classify beer bottles in a german recycling plants
00:00:51
to find defects on medical implants crows
00:00:56
to track players during a volleyball match
00:01:01
or just to find in a completely unsupervised way defects or anomalies on a complex textile patter
00:01:09
these and many more applications are out there still waiting for solution
00:01:15
approaches to uh go about those to use a traditional machine vision but also more modern type
00:01:22
deep learning types of approaches typically and in very
00:01:25
complex highly customised there for costly projects
00:01:29
and they typically also have a very unclear outcome so as a consequence big
00:01:35
companies with big problems and a large resources they can actually a tackle
00:01:40
those kind of problems they can take the risk that can have a couple of engineers to work on it for a couple of years
00:01:46
but there are many smaller companies smaller problems chances out there which are still waiting looking for solutions
00:01:53
and that's basically where our product really sweet comes in it's a comprehensive software package
00:01:59
that provides the basic utilities in order to extract
00:02:03
information from complex data complex images uh like we see here on the screen
00:02:10
it consists of three basic tools the blue tool which
00:02:13
provides a way to track detect and identify
00:02:17
features or objects in images it does that very effectively and very fast
00:02:23
the second tool is i read tool which is a tool to not
00:02:27
only analysts single points but actually regions in which is you know
00:02:31
to to supervise segmentation on images or differences also to find in a
00:02:36
completely unsupervised fashion anomalies or aesthetic defects on surfaces or objects
00:02:42
and finally our green tool which is our general purpose classifier
00:02:47
now those three tools have allowed us in the past two years to us all over two hundred success
00:02:54
starts with the small team that we are and that shows that basically
00:02:58
it it uh they are quite a powerful in their combination
00:03:01
we can actually solve many different problems and we have very specifically targeted
00:03:06
factory automation so basically vision and inspection problems in in a duster context
00:03:13
i've been able to launch a this product in different markets first of all course uh you too huh
00:03:20
let's say of the graphical origin in the watch market but
00:03:23
then also in the mets sect domain in automotive
00:03:27
uh and recently also in the domain of electronics assembly uh over on a shot
00:03:32
food and beverage more recently also an outdoor sort of security and surveillance type of applications
00:03:39
we are proud that we have been able to gain how to you know
00:03:42
get a number of prestigious awards at least have been nominated for those
00:03:46
and the next step what we're looking into is actually do extend our product offering
00:03:51
going in the next step to the next level to take that technology and try to implement
00:03:57
that also embedded platforms are specifically looking into take raw a plaque from from india
00:04:03
what your but go liz that we want to take that technology to make the market has it a bit more in the sense that we
00:04:09
can really do go towards a a true is more camera which will
00:04:13
allow us to furthermore lowered into barriers that we're still fighting today
00:04:19
all of this uh was on by a small a team and which
00:04:23
is located in the centre which is uh in accounting free board
00:04:27
we uh together assembled composes in the domain of competition your signs machine learning a
00:04:33
good um uh but also in visit ordinance sales and we are actually
00:04:36
looking into a growing our team so if you're interested in this technology in