New Blog
This blog post first appeared first on old medium publication (https://medium.com/startuprad-io), and was moved to this blog with the relaunch of our website in summer 2024.
Executive Summary
In this episode of Startuprad.io, we delve into the world of AI and computer vision with our guest Simone Trevisan, a computer vision expert and project manager at Blue Tensor. Simone shares insights into the innovative computer vision platform, Iris, designed for small and medium-sized industries to harness AI without the need for coding expertise. He also discusses Blue Tensor’s collaboration with Huawei to deploy their platform on both on-premise and cloud servers. Join us as we explore the intersection of AI, industry, and digital transformation in this engaging discussion on the German startup scene.
Like What You Hear?
You can tip us here: https://www.buymeacoffee.com/Startuprad.io
Robotics in Quality Assurance: “The aim is to completely eliminate the human manipulating of this kind of objects.”— Simone Trevisan
Follow Us on Social Media!
If you are listening to our content for the 2nd time please consider following us on social media.
The challenges of digitalization in the industry: “We strongly have this kind of tissue made of small and medium-sized industry, and it’s very difficult for a small and medium-sized industry to the value that the digitalization can bring to the processes and also the cloud is seen as a way to take your data outside your industry, your plant, and not as an advantage.”— Simone Trevisan
Our Sponsor Startupraven
A startup’s journey can be a tough one, but it doesn’t have to feel like you’re alone on your quest! Invest in others’ success with us by joining our community of entrepreneurs who are building amazing things every day — no matter how big or small their ideas may seem at first glance.
The best way to find investors and cooperation partners for early-stage startups. Sign up here: https://startupraven.com/
The Evolution of BlueTensor: “We are developing to, let’s say a platform, an artificial intelligence platform, one for the computer vision and industry and the other one for the, raising field of natural language processing with, you know, with Chatchipt AI Became, very, very mainstream, so everyone has very high expectation on it.”— Simone Trevisan
Questions Discussed in the Interview
What unique challenges do small and medium-sized industries face when it comes to adopting AI and digitalization, as discussed in the episode?
How does Blue Tensor’s Iris platform enable small and medium-sized industries to use AI without the need for coding or computer vision expertise?
What advantages and challenges did Simone Trevisan from Blue Tensor highlight in using Huawei’s servers and cloud computing resources for prototyping and deploying new models for computer vision technologies?
How does Blue Tensor’s collaboration with a biotech company illustrate the practical applications of their computer vision platform for high-quality assurance in the industry?
In what ways does the collaboration between Blue Tensor and Huawei address the resource limitations, and machine failures, and enhance data storage and backup capabilities for deploying the Iris platform?
What were the key insights and takeaways from the discussion about the reluctance of SMEs to move from on-premise solutions to the cloud and the need for education about the benefits of digitalization?
How does Simone Trevisan’s professional journey from studying telecommunication engineering to working with Blue Tensor highlight the evolution of the industry towards developing AI solutions tailored for industry applications?
What are the implications of Blue Tensor’s shift from a consulting approach to a product-based business approach, with a focus on developing AI platforms for industry applications?
How does the discussion about the Huawei Connect event shed light on the role of cloud computing, digital transformation, and the need for digitalization in small and medium-sized industries in Germany and Italy?
What insights does Blue Tensor’s experience at the Huawei Connect event provide about the opportunities and challenges in leveraging AI solutions for the industry in the European startup scene?
The Intersection of Telecommunication and Artificial Intelligence: “I put myself into it because I was really fascinated by this kind of topics.”— Simone Trevisan
The Video Podcast is set to go live on Thursday, December 7th, 2023
The Audio Podcast is set to go live on Thursday, December 7th, 2023
You can subscribe to our podcasts here.
Spotify
Apple Podcasts
Castbox
Global Podcast Network (Formerly NY City Podcast Network)
“No Code Computer Vision Platforms: You don’t need to be a developer, you don’t need to be an expert of deep learning and neural network models. You just need to be an expert of your own field of what you are building, what you are generating inside your industry.”— Simone Trevisan
The Guest
In this episode of Startuprad.io, host Jörn “Joe” Menninger welcomes Simone Trevisan, the guest from Blue Tensor, to discuss the innovative work being done in the field of AI solutions for healthcare companies. Simone is a computer vision expert and project manager at Blue Tensor, where he applies his expertise in AI and data-driven models to the development of their computer vision platform, Iris.
With a background in telecommunications engineering from Padua, Simone’s journey has led him to the cutting edge of AI technology, where he and his team are revolutionizing the industry with their no-code computer vision platform designed for small and medium-sized industries to leverage AI without requiring coding or computer vision expertise.
Simone’s deep understanding of both telecommunications engineering and AI has positioned him as an invaluable asset at Blue Tensor, where he spearheads the development of AI platforms tailored for industry applications. His professional journey showcases his transition from a consulting approach to a product-based business approach, demonstrating a proactive and innovative mindset that is driving Blue Tensor’s success.
As the podcast delves into the implications of digital transformation and cloud computing on SMEs in Germany and Italy, Simone’s expertise and experience are vital in providing insightful perspectives on the evolving trends within the industry.
“Computer Vision Platform for Quality Assurance”: “You just have to have the knowledge to annotate your products, but it’s a basic requirement for someone who wants to develop this kind of quality assurance.”— Simone Trevisan
The Startup
Simone Trevisan is a pivotal figure in the success story of Blue Tensor, an Italian AI startup making waves in the industry. Founded by a team of experts, Blue Tensor has achieved significant milestones including successful fundraising efforts, setting them apart from their competition. Their ability to secure substantial investments has allowed the company to accelerate the development of their groundbreaking AI solutions, propelling them to the forefront of the industry. Blue Tensor’s innovative approach to AI for the industry stands out from the competition, offering a no-code computer vision platform called Eyerus. This platform empowers small and medium-sized industries to harness the power of AI without requiring coding or computer vision expertise, a feat that sets them apart in the entrepreneurial landscape.
A testament to its unique achievements, Blue Tensor has successfully transitioned from a consulting approach to a product-based business model, demonstrating its agility and adaptability in the market. Their focus on developing AI platforms tailored for industry applications, including computer vision and natural language processing platforms, further solidifies their position as a trailblazer in the field. With an impressive track record of partnerships and projects, such as implementing a cutting-edge computer vision platform in a biotech company to detect minuscule defects in bone prosthetics, Blue Tensor has set a new standard for high-quality assurance in industrial settings. This, coupled with their collaboration with Huawei to deploy their computer vision platform on both on-premise and cloud servers, showcases Blue Tensor’s commitment to innovation and excellence in the AI industry.
Training AI Models on Defect Detection: “You have to say to the model, you start with the white model, black model With no knowledge. So you have to say him, it, or, you have to say that This is a defect. This is another defect with another name, so you can also, differentiate the detection Of the of the defects, so this is a scratch, this is another kind of defect, this is not uniform color and just with, Let’s say rectangles, boxes around the defect, and You collect a lot of images. You, let’s say, annotate in the sense that you say To the model, this is a defect. This is another defect. This is another kind of defect. You just push all your data inside the model. The model starts his training process.”— Simone Trevisan
Hiring!
You can find open positions with Blue Tensor here: https://bluetensor.ai/en/careers/
The Benefits of Cloud Computing for Computer Vision Systems: “So if we have to do so inside an on-premise server, we have to, buy different storage to have the backup While in the cloud, it’s, reliable because they provide us the backup, with their services, with Huawei cloud, and storage backup services.”— Simone Trevisan
Tune in to our Internet Radio Station here:
Join more than 100.000 monthly listeners and stay ahead of the curve with content from more than a dozen amazing creatives for everything you need to know for Tech Entrepreneurship!
Links / Further Readings / Additional Resources
Feedback
We are always looking for ways to make the show better. Please take this opportunity and share your feedback with us! We would love to hear from YOU!!!
The Interviewer
This interview was conducted by Jörn “Joe” Menninger, startup scout, founder, and host of Startuprad.io. Reach out to him:
Keep Up to Date
Here is our publication calendar: https://calendar.google.com/calendar/embed?src=012i27ak525qi3smgjl8v9tbio%40group.calendar.google.com&ctz=Europe%2FBerlin
Topics Discussed in this Interview
Simone Trevisan, BlueTensor, computer vision, Iris, telecommunications engineering, AI, healthcare, no-code platform, small and medium-sized industries, AI model training, data acquisition, annotation, defects, images, videos, Huawei servers, cloud computing, prototype, deployment, Italian accent, show notes, high-quality assurance, collaborative robot, biotech company, bone prosthetics, contamination risks, implant perfection, on-premise servers, cloud servers, resource limitations, digital transformation, SMEs, education, product-based business approach
Automated Transcript
Narrator [00:00:05]:
Welcome to Startuprad.io, your podcast and YouTube blog covering the German startup scene with news, interviews and live events.
Jörn “Joe” Menninger [00:00:20]:
Hello, and welcome, everybody. This is Joe from Startup Rad dot io, your Startup podcast and YouTube blog from Germany as well as the backing podcast for Startup Dot Radio, the world’s number one Tech entrepreneurship radio. If you’re listening to us for the 2nd or 3rd or more times, please make sure you hit the like and subscribe button wherever you’re watching this and wherever you’re listening to this. You know, guys, we are bringing you contact from Germany, Austria, and Switzerland. But this time, I met somebody at the Huawei Connect in Paris. And, actually, I I made a little Exception because once or twice a year, we have startups from abroad, meaning outside of Germany, Austria, and Switzerland. And this time, we have Simone here from beautiful Italy. Originally from Venetia, I’ve heard.
Simone Trevisan — Blue Tensor [00:01:14]:
Yeah. Hello? Hello? Yeah. It’s a pleasure to be here, of course, and thank you for your invitation.
Jörn “Joe” Menninger [00:01:24]:
Totally my pleasure. We already cleared out the weather because outside, it is snowing for the 1st time this year in Germany. We are expecting something around Ten centimeters here for the Americans, that’s 4 inches. And everybody from Corrales says, 4 inches. We get that in September. But For us here, it’s almost like a blizzard. So that’s why I’m prepared. I have my peppermint tea here.
Jörn “Joe” Menninger [00:01:45]:
You have your tea there and hopefully, You you your heating stays on. We already talked about that as well. As I said, we met at, Huawei Connect. There has been a lot of talk about SME topics, digital transformation cloud, NII. And actually, I also tried the, AI Pango model. I was pretty impressed because I was talking to it in German and really fast. And, my German name is Jor, j o with the 2 dots on it, r n. And without a hinge, it translated it.
Jörn “Joe” Menninger [00:02:21]:
And I was Really, really impressed because there’s a lot of big cloud providers who don’t get this. What What did you take from the event? What was some of the, highlights, some of the ideas you took, you took away with you?
Simone Trevisan — Blue Tensor [00:02:38]:
Oh, yes. It was very amazing event. We have, the chance to to see a lot of, solutions That are moving from the, on prem solution to the cloud approach to actually exploit The the power of, cloud computing, that is something that we can Start to think about because, we need more and more computational power so the cloud is a solution to balance, not to waste resources because we can’t just have this kind of virtual, Environment on the cloud, that are scalable, that we can, yes, use on demand and Not waste, of course, resources that maybe we we don’t need in their full, And, also, there is a very strong, approach, a very strong idea on digitalization because, yes, the the consumer are very, a very on the edge with the new technology. They are very they use a lot of new technology technologies. Everyone is using ChatCBT, these are cutting edge natural language processing technology, but we forget that maybe the industry is the, especially the small and medium, industries are still Behind, they’re using a lot of, more traditional, a lot of, handmade, processes in their production lines. So this I I saw a very strong, direction, a very strong, I mean, intention to Bring the the digitalization to the to the industry.
Jörn “Joe” Menninger [00:04:50]:
Actually, I have to admit just recently, I saw some SMEs still using floppy disk, moving it from computer to computer. So there is a lot of improvement to be done here.
Simone Trevisan — Blue Tensor [00:05:03]:
Yes. What’s up?
Jörn “Joe” Menninger [00:05:05]:
Not too recently, I saw SMEs here in Germany moving data with floppy disk from a to b around. So I think there’s a lot to do as you’ve been talking about, the processing power of cloud. Personal, personal feeling is that there’s a lot of digitalization to be done, but also moving on premise stuff Into the cloud where I have to admit a few of the, very long time entrepreneurs are still a little bit frightened of, but I get the feeling that Germany, especially the SMEs, are a little bit behind the curve here. How is your, like, gut feeling in Italy right now?
Simone Trevisan — Blue Tensor [00:05:48]:
Yes. Also, in Italy, I can say that, and in the Industry the industry environment in Italy is mainly small and medium size, and they use They still use a more artisan approach when they in their production processes. So, Yes. We are I can say that we are behind it. Also, say behind German if, I mean, we we strongly Have this kind of tissue made of small and medium sized industry, and it’s very difficult for a small and medium sized industry to the value that the digitalization can bring to the, 3 day processes and also the cloud is seen as a, I mean, as a way to take your alpha outside your your industry, your plant, and not, as an advantage. So this is, a very point that we have to work a lot because cloud is Not meaning taking your data and, moving them outside your your plant is, made moving them in a more secure, storing system, A more, reliable storage system you can access to much more, power powerful, processing, computing, and, yeah, this the point where we have to I mean, we have to point the most when we talk to our clients, to our customers.
örn “Joe” Menninger [00:07:46]:
Just to make sure, if you’re talking about secure storage, you’re talking about digital storage and not stone tablets. Right?
Simone Trevisan — Blue Tensor [00:07:54]:
Oh, yes.
Jörn “Joe” Menninger [00:07:56]:
Just to make sure, before we get a little bit into what you guys are doing, I I’ve seen you you did some very impressive stuff. I have to admit, I don’t understand a lot because, My Italian is, so to say, nonexistent. But but but but I do get that you studied at Padua? And then you did, some type of internship, then you’ve been with an information technology company, And then you already started at Plutensor. Can you take us a little bit along the journey for, like, 1 or 2 minutes
Simone Trevisan — Blue Tensor [00:08:32]:
and we Of course.
Jörn “Joe” Menninger [00:08:33]:
There so People understand from from who whom they are getting the good advice.
Simone Trevisan — Blue Tensor [00:08:39]:
Yes. Of course. Yes. I graduated At the University of Padua in telecommunication engineering. And during my master degree, I had a internship in a consulting company of inter of information technology In Milan. And then when I graduated at the university, I took maybe a few months, to go around the world. I stayed in Thailand. Yeah, I enjoyed the time to just organize my my ideas on my future, then I came back and then moved To train to a software working in Blue Tensor.
Simone Trevisan — Blue Tensor [00:09:31]:
When I moved to Blue Tensor and I started my my journey, I, it it was still because BlueTensor was born in 2018 as a start up. Now I we are almost at the end of our startup journey. But, Yes. We started with the strong consulting approach, and our, aim is to, Develop, tailor made, AI solution for the industry. But in the last years, we started to Synthesize our experience in products. So we are developing to, Let’s say platform, artificial intelligence platform, one for the computer vision and industry and the other one for the, raising field of natural language processing with, you know, with Chatchipt AI Became, very, very mainstream, so everyone has very high expectation on it. So We developed these 2 products, and we are moving from a strong consulting approach to, let’s say product based business approach. Yes.
Simone Trevisan — Blue Tensor [00:11:03]:
I started as developer as a computer vision expert in in BlueTensor. Yes. Now I’m when we started the, Our computer vision platform, Iris, that that’s the name of the the name of the of the platform. I started as a project manager of this platform, and I’m very proud on How it’s evolving, how we are developing, and also how we are keeping it at the edge of the computer vision technologies.
Jörn “Joe” Menninger [00:11:41]:
Computer vision is a pretty amazing topic. We’ll we’ll get right To it. But 1 question, when when you talked about, telecommunications engineer, I had in the back of my mind The the the main question. So every time the satellite dish of one of your relatives doesn’t work, they call you. Right?
Simone Trevisan — Blue Tensor [00:12:05]:
Yes and no? No. Actually, I chose a very, let’s say, weird Curriculum for my telecommunication, degree, I attend I, of course, attended a lot of Networks, wireless networks, 5 g and mobile networks courses, Signal processing, so this kind of I mean, mainly, telecommunication fields, But also, it was the start, I mean, of the AI. So also the university, The university started a lot of courses about artificial intelligence or data driven models. I put myself into into it because I was really fascinated by this kind of topics. Also, I attended course about the application of these information technologies, both AI and data driven models and signals processing into health care application. I, I’m actually, let’s say, information health care information technology Expert. But, yeah, this is my my telecommunication curriculum. So It’s, yes, about telecommunication telecommunication, about signal processing, but applied to data driven models In this case, health care.
Jörn “Joe” Menninger [00:13:52]:
Mhmm. You are talking about AI no code visualization. You you your tool is called I r Us. Right?
Simone Trevisan — Blue Tensor [00:14:06]:
Iris. Iris. Yeah. Iris. Name of the platform. Exactly.
Jörn “Joe” Menninger [00:14:10]:
Spelled like irus.ai. And and can you first before we get into a little bit specific stuff, can you tell us what an AI no code visualization actually is because in in the brain of most people right now, there there’s the buzzword alarm going off. A lots of buzzwords, no meaning behind it.
Simone Trevisan — Blue Tensor [00:14:35]:
Can can
Jörn “Joe” Menninger [00:14:35]:
can you put a little bit meaning behind this for all of us?
Simone Trevisan — Blue Tensor [00:14:40]:
Oh, yes. Iris is our computer vision platform and has been has been designed To, not require any coding expertise, any computer vision Expertise, it’s just something that allows the industry, especially this small and medium size Industry to, embrace, to leverage the power of AI without the Need of high budget or high investment on maybe a computer vision team or data scientist team. So this is the meaning of no code, computer vision platforms. You know, you don’t need to be a A developer, you don’t need to be, an expert of, deep learning and neural network models. You just need to be an expert of your own field of what you are of your product or what you are building what you are generating inside your, inside your industry. So this is the only requirement that you have to start using our platform.
Jörn “Joe” Menninger [00:15:59]:
Mhmm. Mhmm. I see. And and now Let’s talk a little bit about it’s basically an AI. I don’t have a degree in engineering. Okay? So so Cut me some Slack here. And could could I put it in very layman terms that your platform is an AI you can train that works with pictures.
Simone Trevisan — Blue Tensor [00:16:28]:
Yes.
Jörn “Joe” Menninger [00:16:30]:
Or almost an engineer. Correct.
Simone Trevisan — Blue Tensor [00:16:34]:
Yes. It’s a computer vision platform that, of course, works with images or also video. The point is that you can, start the the point the development of an entire computer vision system. So starting from the, data acquisition, Data annotation, AI model training that is one of the, I mean, the annotation and the training are one of the most difficult part of, computer vision System AI system, development, and then the one of the main feature of our platform Is that you can with our flow technology that we we develop, you can, take the model that you developed, put it inside a workflow, and deploy it inside your, production line with, of course, no need of code, no need of, expertise. You just have your computer video system For, let’s say, one of the, main field of application is the quality assurance, so you can start acquiring Acquiring your images to Iris, you annotate them through Iris, using also these, technology of Data notation that allows you to be very, very fast. Also, it supports you with some smart tools And then train the model. And finally, you can just deploy your model in your workflow, and you have your computer vision working with without, again, writing a single line of code and without having the knowledge, behind the computer vision model. So you just have to, have the knowledge, to annotate, I mean, your products, but, I mean, it’s basic requirement for, for someone who Who wants to to develop this kind of quality assurance.
Jörn “Joe” Menninger [00:18:55]:
Mhmm. And 1 question. When you’re talking about annotation, it’s basically You have what I have in mind. You have a picture. You put, some circles around several pieces, and you then tell the AI, for example, That is a good piece. There’s something broken off, and we want to have it look like that. Something along those simple lines?
Simone Trevisan — Blue Tensor [00:19:16]:
Oh, yes. It’s something like you have your images
Jörn “Joe” Menninger [00:19:20]:
Mhmm.
Simone Trevisan — Blue Tensor [00:19:21]:
And you Have to say to to the model, you start with the, white model, black model With no knowledge. So you have to say him, it, or, you have to say that This is a defect. This is another defect with another name, so you can also, differentiate the detection Of the of the defects, so this is a scratch, this is another kind of defect, this is not uniform color and just with, Let’s say rectangles, boxes around the the defect, and You collect a lot of images. You, let’s say, annotate in the sense that you say To the model, this is a defect. This is another defect. This is another kind of defect. You just push all your data inside the, the model. The model start his training process.
Simone Trevisan — Blue Tensor [00:20:32]:
And at the end, you you have a model that is able and, very, very specialized in recognizing your defect Or the defect that you provided, and not only in the images, of course, that you provided because, This is very simple, but, the strength of the AI and what differentiate, the AI models from traditional approaches Is that the AI models are very, very good in generalizing the defects. So they just extracted features of the defect. They learn the feature of what is a defect, And then they start looking for the defect in new images, and new pieces that, he never he never seen before.
Jörn “Joe” Menninger [00:21:25]:
So so, basically, you have to, do some training on the job as you would like a human person, in quality insurance, but you have to dumb it down below human level as you would have speaking with a very dim with a toaster, And you train it over time that it gets as good or even better as your best quality assurance people.
Simone Trevisan — Blue Tensor [00:21:48]:
Oh, it’s, You start with, you start we start with a baseline, but, that is more or less, at the same level of a human level, quality assurance. And then, with our models, we, Let’s say we start training the models, and you can think about the models like babies. So they cannot do a lot of things. They just Can learn to make few things, but can learn to make them very, very, very good. So if you show them the defect, They start, learning how the device is made. And at the end of the process, they are, In most of the cases, they are much better than human, level of quality assurance, control. And then, for more reasons, one is the bias. The the human brain is strongly biased.
Simone Trevisan — Blue Tensor [00:22:55]:
Also, we had Scenarios where the machine that is able to extract feature from the image is able to detect defects that, were not detected by the human. It was very difficult to see, by the humans. So, we can reach a very high performance with this kind of models.
Jörn “Joe” Menninger [00:23:22]:
Mhmm. I see. And and, basically, Well, you talked about the babies. I I know how you feel about your models. Let’s talk briefly about how you actually apply something Before we before we start the, interview, you talked about an implementation you did in a biotech company Producing bone prosthetics. So how you’re actually using that? And then we can get a little bit back to to the topic that Connected us by help of Huawei. But first, we want to have, like, a real world example how you could do that because a lot of our audience, that was pretty abstract what we’ve been doing. But if they do get, like, a real feeling on, what it can actually do, They start getting ideas like maybe today, maybe tomorrow, maybe after 2 beer, maybe after a dozen beers, doesn’t matter.
Simone Trevisan — Blue Tensor [00:24:16]:
Okay. Yes. Of course. We, developed a very interesting project, in the biotech technology. As you said, there was this, prosthetics, bone prosthet hip prosthesis is, of course, made of a metal part and another part that It’s inserted inside Gong. And, the problem is that, we have to reduce, the risk of of contamination of this part, are products that needs a very, very high level Of Cortic Assurance. And, For for
Jörn “Joe” Menninger [00:25:03]:
the very simple reason, it it goes Into a human body. So no yeah. No person can really touch it with a finger. Best nobody in the room. Right?
Simone Trevisan — Blue Tensor [00:25:12]:
Yes. Yes. Of course. So the quality assurance Has to be made inside a clean room. So, there are solution where the, A human can go inside and use the gloves and do the quality assurance, but, the aim is to Completely eliminate the human manipulating of this kind of objects. And So we used, Cobot, a collaborative robot. So we also integrated this kind of technology that take their prosthesis, Bring it, in the front of the cameras so we can acquire images with different lights, Then it put it down, it take with the other side to control also the metal part. The metal part is the part that goes With the metal sphere that goes, in the joint, And then we we control through Iris that the piece does doesn’t contain any defect.
Simone Trevisan — Blue Tensor [00:26:26]:
That is very important. They are very, very they have very, very high standard. The pieces must not contain any kind of defect and that kind of color, different color for Because it means that it has not been, let’s say elaborated correctly by the the human, the machine, the prob Yeah. The That’s it.
Jörn “Joe” Menninger [00:26:56]:
It has to Fulfill higher standards and the highly reliable, machines. Because if if one of those implants is broken, You cannot just send in a senior senior engineer with a monkey wrench with a full cup of coffee. It doesn’t work because you also After you surgery again and per the patient has to Of course. Occur additional risks. So the they have very high standards. Right? And you basically
Simone Trevisan — Blue Tensor [00:27:22]:
Yes. And also you have very, very small defects. That means that maybe the, clinician cannot maybe, see, those defects when with his, eyes and when you implant it inside the body and you start Understanding that there is something wrong with it, it’s too late because you have to, remove it and change it. And it’s Not something that someone wants to go away.
Jörn “Joe” Menninger [00:27:52]:
Are we talking about something like microfissures here?
Simone Trevisan — Blue Tensor [00:27:55]:
I’m sorry. What?
Jörn “Joe” Menninger [00:27:57]:
Like microfissures, like very, very tiny, fissures, like broke, breaks, like very tiny that you can usually only see.
Simone Trevisan — Blue Tensor [00:28:05]:
Yes. Very small scratches, maybe. We when we started with Blue Tensor, everyone was happy with Millimeters defects. Now they are asking us for a smaller and smaller defect. Now we are in the range of fraction of millimeters. We have to detect very, very small, imperfections in the surface of this kind of Products.
Jörn “Joe” Menninger [00:28:33]:
Mhmm. My understanding is, as you said, you’re already, working there with the robot. So, basically, Robert has taken it, has taken the pictures. It gets sent to the Huawei cloud to your model. It comes back and it says, oh, good or no good, and then it proceeds Without a human ever touching it.
Simone Trevisan — Blue Tensor [00:28:51]:
Yes. Of course. Yes. Yes. The room is closed, so So nobody can enter inside the room, before the process has ended. So, the Cobalt is able to For all the manipulation that we need to perform the quality assurance.
Jörn “Joe” Menninger [00:29:11]:
Mhmm. I see. And now let let us go a little bit back in in order to, like, close the loop. We talked about, Huawei Connect. How did those guys at Huawei help you? Because I I could I could totally see how the person in charge of the, acceleration program could help you. He’s a very talkative, very in energy full guy. But how did the company help you?
Simone Trevisan — Blue Tensor [00:29:36]:
Oh, yes. We started our collaboration with with Huawei with the With the program for we applied for for a program for for a acceleration program, where we, proposed That our computer vision platform, Iris, that now is installed, is deployed inside On prem is servers, so the server is inside the company, inside the company network. So, it’s Okay. Because everyone everything stays inside. Also, we have limitation on computation of, resources. We have a limitation on easy is easy of access because, of course, we use the Standard firewall of our machines, but, of course, something somewhat, Sometimes not enough. And, also, when the machine fails, It’s a problem because we have a single point of failure if we have a single machine deployed inside the the company. So our proposal is to, move or not move completely, but to extend our offer And move I was not only on premise, but also as a service on cloud.
Simone Trevisan — Blue Tensor [00:31:05]:
So we can provide, Our unique features, our, computer vision platform, and the ability to Develop on your own your extremely customized computer vision system to the cloud without the need to buy machines so that are kinda expensive And also with the reliability of the cloud, the easy access of the cloud that you can access wherever you are, and you can You can use more sophisticated firewall system, in that the cloud services as more way are providing us, and we can also exploit data storage because, as I said, they are asking us to Detect smaller and smaller defects. That means that we have to, generate bigger and bigger images To detect very, very small defects, so we need more computational resource to process this kind of Images that are very big and also have to store them and we have to store them in a reliable way. So if we have to do so inside an on prem server, we have to, buy different storage to have the backup While in the cloud, it’s, reliable because they provide us the backup, with their services, with Huawei cloud, and storage backup services.
Jörn “Joe” Menninger [00:32:49]:
Do they also help you with the with the some of the models?
Simone Trevisan — Blue Tensor [00:32:54]:
We don’t use the models, or we actually use the computational power of the Huawei servers With the, to train our models and to prototype our new models because also we want to keep, Iris at the cutting edge of their computer vision technologies. So we are always Trying to find new models that can perform better or, models more suitable for Some scenarios. So we use the, let’s say, Kinda limitless power of the of the cloud because it can provide you a lot of Service, a lot of GPUs, a lot of, computing resources to prototype, Deploy new models and, of course, train models. That is another very big time, a very, computer resource Expensive, task. Mhmm.
Jörn “Joe” Menninger [00:34:06]:
Great. Simona, I would let you go because usually we talk around 25 minutes now. We’re recording for almost 35 minutes, and I can only tell your interview will be a big success, especially with ladies because your Italian Italian accent is amazing.
Simone Trevisan — Blue Tensor [00:34:24]:
Thank you.
Jörn “Joe” Menninger [00:34:26]:
Also, the story was pretty interesting. We’d link down here in the show notes a lot of additional links. We get additional information whether you’re watching this or listening to this. You can go down here in the show notes or to blog.startuprad.io, and there you’ll find the show notes for the respective episode. Simona, the Only thing for me to say is. Great. It was a pleasure talking to me. Thank you.
Jörn “Joe” Menninger [00:34:56]:
Bye bye.
Simone Trevisan — Blue Tensor [00:34:57]:
Thank you. Bye.
Narrator [00:35:03]:
That’s all, folks. Find more news, streams, events, and interviews at www.startrad.io. Remember, sharing is caring.
Comments