Who are EVT and how do you fit into Machine Vision?
EVT is mainly a software company and our USP is a universally usable software with a fast time to market access. The software is graphically programmed and has a full choice of hardware platforms, so you can use smart cameras and embedded systems for PC's with Windows or Linux. EVT fits into the vision area, because we make vision software to do analysis in 2D, 3D, 1D. We also have the latest generation of analysis which is thermal imaging and hyper-spectral. This offers new applications fields in the machine vision environment for us.
What industry is your primary focus?
As we offer a software for all applications, we fit into nearly all industries. And with our integrators the software is used in wide area machine vision applications, which makes it easy for us to not only focus on only few solutions. We’re in the fields of, automotive, a major industry in Germany, electronics, semi-conductor, food & beverage, rubber, woodworking and some others. The focus is defined by our main customer base but also partly changes according to new inventions on the market and new customers. Sometimes the focus is in automotive industry, sometimes its semi-conductor depending on the market. For example, to the Chinese market we cater to the electronic industry due to the manufacturing of smartphone parts. If we would integrate more in Finland, we would probably concentrate more on woodworking for example. It depends on the countries where the integrators are and where they have their knowledge producing machines in different markets for the end customers.
What have you found that the applications are most suited for your out-of-the-box applications?
As we offer the software for all applications, the question what is the “most suited” application is tricky. Of course, if you use our software with smart cameras and very easy solutions for reading, counting, color analyzing and matching. If you have plans to do PC solutions, it has changed in the last 3 years from 1D to 2D and now 3D, through this growing area with a lot of requirements. Some new customers really have great need for 3D solutions. There are also cases dating several years back, when it was not possible to deliver them solution they needed, but which is possible now with new technology as well as the software and CPU. Now we can offer these customers solutions all in 3D. Things slightly change from time to time and I'm sure next year our next biggest market or one of the bigger markets we will see is thermal imaging. This is not passive thermal imaging which can be done very easily like normal Gray scale, no I think that impulse tomography or lock-in tomography will be a real big market.
There are so many options or requirements from customers that you can solve now and some that you can’t because the technology is still advancing, so we will be able to solve those sometime in the future.
In terms of the hardware that's used are there any key customers of yours that you can talk about and give some application examples?
There is for example an industrial solution in the beverage industry, which uses some of our 3D sensors. They are also still using the 2D sensors but with 3D sensors they can now inspect the pressuring of bottles and of cans. This wasn't possible earlier, because the sensors were not fast enough, or the sensors were fast enough but too expensive to produce.
Recently there are also low-priced sensors available, which are still fast enough. This means that they can be used by the industries for their machines. Another interesting thing is the food industry which started a few years ago and is still growing. We are now inspecting all kind of convenience food in the production line. It is not the same as inspecting fresh fruit, where a rotten tomato means real trouble. However, the customer wants to see when they buy pizza, that they have the same amount of a salami on the pizza. Or if you buy baked bread you want it all to have the same shape and it should look nice, so this inspection is an excellent application. It can be used for canned food during the production process, for example if your green beans are all okay. Some products can only be monitored with vision and in this area, we have some nice applications.
We also work in the pharmaceutical industry, based on the regulations and the requirements from the different governing bodies, to prove that your production 100% conforms with regulations as well as quality control checks. The bin picking applications are still growing, and the 3D solutions fit the environment. Therefore, we have integration for robots to make it easier to use the 3D image capturing with the robot bin-picking. Another example is from the semiconductor industries to inspect the BGAs, if they are okay.
Have you found that with your work here you've been able to see that people are developing the software a lot faster than the hardware?
We found that within the Machine Vision people are looking for more developments in software to back up their existing sensors. Or they are looking for new sensors and stick with their old software. There are, of course, many people which have old systems, but still want to maintain their machine. But when the old machine is no longer working, they need a replacement and based on our software technology, we can solve the task and find a camera which fits in their housing. Therefore, they simply can maintain their old equipment with the new technology like DV cameras replacing old analogue camera. In the machine this is the ITN for the machine.
In some cases, the software is old because you find a lot of replacement hardware, but you need software which allows you to make the solution very fast. So, you can buy a new system to use the special software for that but it's very costly.
What are you looking to develop further with EVT?
I just mentioned thermal imaging is a growing market, so we invest a lot in thermal imaging. We have also started to develop artificial intelligence, deep learning or machine learning networks to solve tasks which are nearly impossible to solve with standard algorithms. It can be described as compositional flow. We have these AI systems and we use the standard engine extensive flow, which is uses for all the other applications. So, the training is different but inside it is the same. You can solve difficult tasks more easily if it's split into more tasks.
This will be one of the major developments for next year additionally to the new cameras like the integration of hyper-spectral cameras. Although a hyper-spectral camera only delivers groups of images it is easy to feed it into the deep learning network and get the result.