Intopii

Today, we released a shiny new beta of Into, the machine vision and AI platform of the future. What could possibly be a better way of spending your Christmas than coding parallel image processing applications in C++ or JavaScript? This is the first version publicly available to anyone without registration. Happy coding!

(Posted by Topi)

Kide Quality Guard just got a little brother, Kide Defect Guard. It is a flaw detector for sheets, boards and webs, automating man-made inspection with an affordable price tag. Kide Defect Guard brings the benefits of powerful machine vision systems to applications where the usual cost of such systems cannot be justified.

It's really compact.

Kide Defect Guard features a camera, digital I/O and sophisticated computer vision techniques in a package that fits in your pocket, making it one of the smallest surface inspection systems ever made.

Turns defects into digital outputs.

Camera parameters, defect detection and classification are set up with an intuitive configuration software through a local network. Defects can be classified based on their appearance and size. Intopii's self-learning image classification program, Pinta is used to easily build classifiers. Once set up, Kide Defect Guard works as a sensor; it signals about detected defects through its integrated digital outputs. Real-time user interfaces and reports are available as options.

Fits to a wide range of applications and industries.

Kide Defect Guard works with plastic, non-wovens, textiles, and construction boards, among other materials. We are currently testing the concept on a plywood line. More information about Kide Defect Guard available at http://intopii.com/en/products/kide/overview.

(Posted by Janne)

A public beta of Into 2 is out after six months of more or less intensive work. The new version comes with loads of performance and stability improvements, a new object recognition framework, MSVC support, experimental QML support and much more. Go get a free copy and see yourself!

(Posted by Topi)

The first Kide Inspector at Wipak has now inspected the plastic line for almost two years. Satisfied with the proven performance and benefits of Kide, Wipak just coudn't resists the temptation of replacing the existing machine vision system with a new Kide system also on the another production line. Well, we didn't have anything against it either.

We even smoothed their path to investing decision by recommending our rental concept.

In Kide rental consept the Kide Inspector is firstly rented for customer and taken into production use. During the rental period the customer get familiar not only with Kide but also with their own product quality. This helps both of us to tune Kide's features and classification accuracy at the highest possible level. Often Kide gives valuable feedback to customer's operators for process control purposes already during its firsts working days. A first steps for enhanced process control and continuosly better product quality are usually taken before we even left the factory floor. Thus, Kide starts paying the investment back even before customer has invested it to its own balance.

With hard facts and proven experiences the final investment decision is easy to make. That's what Kide rental is for.

Well, Wipak get hooked to you rental model and on 6th October we announced at Twitter that Wipak has rented their second Kide System. Today, we are happy to announce that Wipak declared of investing the rented Kide Sheet and Web Inspector.

(Posted by Janne)

Intopii's employees have started celebrating Christmas. During the Christmas time our office is closed. Most of us will work every now and then during the week from 23. December to 3rd January 2011. Therefore, we ask your patience as responding to your e-mails might take a bit longer than normally. On 3rd January we'll be back making things think and helping you to do so.

Thank you for the year 2010!

Intopii wishes You a Merry Christmas and Happy New Year 2011!

(Posted by Janne)

We are happy to announce that a new version of our Automated Image Analysis SDK, Pinta (intopii.com/pinta), is out and available for download at intopii.com/en/products/pinta/download. Running now also on Mac, beta-5 comes with a load of improvements and new features.

Pinta is the fast and simple way to create high-performance applications that know what's in the image. It makes your applications able to classify, segment, and recognize images and videos based on their contents. Pinta comes with Trainer for creating analyzers and an API for integrating them with your software.

2.0-beta-5 boasts significant new features and improvements that make you able to create more powerful autonomous image analyzers with even less effort and time than before. These include newly written image cache that radically shortens learning time and reduces memory footprint when processing large amounts of image data, color correlogram features (see bit.ly/cuKk71 for an explanation) that enable your analyzers to understand spatial distribution of colors in the image (which color histogram, for example, does not allow), and feature combiner that allows analyzing multiple image features (such as texture and histograms) at the same time. The user interface of Pinta Trainer is now yet more intuitive and beautiful. We also trashed some features on the labeling side that made using Pinta more complicated while adding practically no value.

But what we are maybe the most excited about is the fact that Pinta now runs, in addition to Linux and Windows, also on Mac! We hope Mac developers will welcome the possibility to make their applications understand images using the unique, simple, and quick way Pinta offers.

There's one more thing we'd like to mention. Besides being a great product on its own, we think Pinta is a premium example of what can be created with Into - the most potent software framework for building cross-platform applications that leverage on computer vision and machine intelligence technologies. Into is currently in limited alpha phase but available on request for selected developers. We welcome applications at intopii.com/into.

That said, don't forget to get your hands on Pinta at intopii.com/en/products/pinta/download and get creative!

(Posted by Heikki)

Into is a cross-platform machine intelligence application framework written in C++. It's open source, licensed under the Affero GPL v3. Into provides a different, super-fast way to build cutting-edge, high-performance applications for image analysis, machine vision, pattern recognition, and artificial intelligence.

It features a layered API and more than 20 fully interoperable plug-in modules for accessing image and data sources, powerful feature extractors, classifiers, neural networks, and much more. Into provides more than a typical software library or API. It features an innovative execution engine, Ydin, that makes it easy to create dynamic programs that automatically run in parallel. Into lets you create complex, distributed applications with unprecedented ease and speed. Close intergration with Qt makes creating beautiful user interfaces a snap.

Commercial licensing options will be made available later this year.

Sign up and download Into at http://intopii.com/into

(Posted by Janne)

In the spring, we installed Kide to a customer's sheeting line. The ramp-up phase was somewhat laborous due to problems in installation and the wide variety of different sheets the mill was producing. This fall, everything seemed to be finally OK. We had been able to train the system with almost all product flavors, and it seemed to do a decent job. The system controlled the sheet sorter just fine and had been running steady for a couple of months. So we decided it is about time to roll in our acceptance procedure and declare the project formally closed.

The first 200 sheets went fine, and we were at our performance target of 95% correct classifications compared to human evaluators. Then, all of a sudden, the system started to reject almost half of the sheets for edge cracks. None of us had a clue of what was happening. Needless to say, the customer was not too convinced. Neither were we, as nothing, really nothing should ever cause such behavior.

Back in the office we reran the experiment with a simulator back and forth but could not reproduce the problem. No matter what distortions we added to the images the system just didn't fail. Then, suddenly, we saw an incorrectly classified sheet. The system reported an edge crack, but the simulated sheet was fine, just with a bit of artificial distortions. "A bit" may however be an understatement. When we tracked the failure down to the code, we found that four preconditions must be met simultaneously: 1) part of the image background must be lighter than the rest (there was simulated white dust) 2) the edge of the sheet must be close to this area (we simulated a sidewise moving product) 3) contrast must be bad (the simulated sheet was black) 4) there must be a large dark defect on the sheet (we added simulated dirt). If all these occured simultaneously, our real-time edge tracker gave unreliable estimates for a while. Happily, the problem was quite easily solved once found.

What did we learn, then? At least the fact that in industry, halting a production line or rejecting good material is not an option. An inspection system must work at a nearly 100% performance level 24/7. A single unexplained failure will render the system unreliable and cost an unacceptable amount of money. The challenge is that there is no way to simulate everything in a lab. It takes a heap of tuning on mill floors to get a product to a level that is required for serious applications such as on-line control of a production line.

(Posted by Topi)

Pinta 2.0.0-beta-4 is out. Since beta-3, Pinta has been able to analyze and classify images based on their colors. This release closes a couple of bugs and finishes undo/redo functionality. The C API has recently been tested in a heavy industrial application running at 100fps. Have a look at some other Pinta applications at http://intopii.fi/en/products/pinta/applications!

(Posted by Topi)

Pinta beta-3 was finally released today. The new release enables you to analyze and classify images based on their colors, thanks to the addition of color histograms and color percentile features. Several other improvements, such as an enhanced UI, are included as well. Please check http://intopii.fi/en/products/pinta/download  for the details, grab your copy, and see for yourself!

(Posted by Heikki)

Pinta beta-3 is scheduled to be released on next Monday, 9.11.2009. The new release will enable you to analyze and classify images based on their colors, thanks to the addition of color features. Several other improvements will be included as well. Please check http://intopii.fi/en/products/pinta/download  for the details.

(Posted by Heikki)

Pinta 2.0.0 beta-2 is out. The new image splitter option makes it better for segmenting purposes.

• A bug in Pinta API resulting in side-by-side configuration errors with certain Windows versions was fixed.
• Option to split images into smaller pieces to be classified was added to preprocessing options.
• Option to adjust how thoroughly/quickly the classifier is trained was added to classifier options.

Pinta image classifier gets you hooked.Try it!

(Posted by Janne)

Finnish Candidates for European Parliament try to show their faces just everywhere. The main problem of this election seems to be that voters do not know candidates, not to mention the fact that most of the voters do not know voting date, either. Naama Candidate Finder, released by Intopii, is giving helping hand to the voters and candidates. Naama matches faces and tells who to vote.

Naama (Finnish word for mug or face) Candidate Finder provides voters with an easy, entertaining, and fast way to select suitable candidates in the European Parliament elections on June 2009.

The Finder is very simple to use: the users only need to upload their pictures. As a result, Naama shows them a list of the most appropriate candidates or so-called face twins. When compared to many traditional candidate finding services Naama has one huge advantage: instead of forcing the potential voter to go through an exhausting set of questions it gives an instant result.

Is somebody really voting based on facial look-alikeness?

The research done and published in USA has shown that voters are likely to vote the candidate who mostly resembles themselves. The look-alikeness effect is the more effective when candidates are not well-known from other context. The link to the research is available at www.naama.fi.
 
Naama was published also during the Finnish Parliament Election on 2007. It gained huge popularity in Finland and abroad. During a three week period over 167 000 voter searched their candidate with the help of Naama. The user rate equals to 4 % of all persons who had right to vote.  Furthermore, according to Naama’s poll, 25% of the visitors say that they will be voting by face value alone.

Naama is developed using the Into, a software platform targeted at solving applications that need intelligent reasoning. The platform can be used to analyze any kind of digital data, and it is being widely used in demanding industrial image analysis applications.

(Posted by Janne)

Dear customers, suppliers, and occasional visitors! Some of you may have been waiting for our annual announcement about a new delivery of cutting edge pattern recognition technology to Santa Claus. Well, here it is ...and that's all about it. Sorry folks, we cannot say more - it's top secret this time. But rest assured, it's something very fancy - as usual.

Anyway, we'd like to take this opportunity to thank you all for this year and wish you a merry and relaxing Christmas and happy New Year! 

(Posted by Heikki) 

We are happy to announce that we will deliver advanced pattern recognition software to Sesca Group for an application in copper refinement process. The software will be used to recognize cast copper parts. 

The solution is based on your pattern recognition database, previously used in the famous face matching service, Naama.

(Posted by Heikki)

We just couldn't resist the temptation...

(Posted by Heikki)

Thanks to everybody who visited our stand at Machine Vision Day at Tekniikka 2008 fair in Jyväskylä! Some interesting opportunities emerged, and we'll start working on them very soon.

Tekniikka 2008 was the last one for us as exhibitors this year. But the next year comes flying and it's highly probable that we will be present at Automaatio 09 at Helsinki Fair Centre, at least. We'll get back to you on that later.

(Posted by Heikki)

P.S. As usual, our stand stood out among the others, while being the only one where you could comfortably sit - and that's on couches, not some shaky bar stools :) Rest assured that we will try to make your experience even more comfortable next time.

I found it unappealing to travel to Saariselkä (it is over 460 km from here) just to switch a single bit in BIOS settings, but there seemed to be no remedy. No, wait. Hadn't I heard about something called /dev/nvram? I had to grab my laptop and give it a quick check at about 11:30 pm.

Next day at work. Remote connection to the Otos box at Saariselkä. Save the CMOS dump. Take similar dumps of all our computers to find one with the same binary layout. Only one seemed to match, but that should be enough. Now find out the bit that needs to be changed. Turn on the computer, make the change to its power management settings, take the CMOS dump again and see what was changed. Great, I got you.

When the CMOS battery goes low, the memory will soon get garbled. That is why the boot-up software places a checksum somewhere into the memory. I knew it was a CRC16, but had no clue where in the memory it was placed. Worse yet, I had no idea which bytes it was actually calculated from. So I wrote a short C program that tried all reasonable combinations. Finishing the job was easy: take the original dump, change the autoboot bit, recalculate the checksum, and store it, copy the modified memory dump back to the computer at Saariselkä, and reboot the machine.

Finally, with all tests, it had taken me four hours to change a single bit in a computer's memory! Some people calculate productivity based on the number of lines written...

(Posted by Topi)

Addendum 2008-09-30: It works! Heikki returned from his trip to Nuorgam and reported that he had succesfully tested the machine. He also installed an UPS there.

Intopii will be one of the exhibitors at Tekniikka 2008 trade fair in Jyväskylä Paviljonki. We will present our products in the second floor lobby, in connection with the Machine Vision Day on 2.10.2008.

More information about Machine Vision Day available in Finnish at http://www.jamk.fi/konenakojkl/
More information Tekniikka 2008 available at http://www.jklpaviljonki.fi/tekniikka2008/

See you there!

As some of you, our dear visitors, surely have already noticed, we have overhauled our website. Officially welcome! It's nice to have you here.

Our old site was admittedly quite thin content-wise and technically complicated to manage and keep updated. Now we have the site running on a real content management system that has already made managing it a lot easier and faster. Thanks go to Olli for all the server side stuff and much of the design, and everybody else for comments and other contributions to the content of the new site.

Speaking of content, there is already much more of it than ever before on Intopii website and we'll be adding more. The goal is to keep our site always full of relevant information about our products, services, projects, and so on. Just subscribe to our news feed and/or news letter to monitor changes and keep yourself informed!

What is more, we thought that it was time to update our imagery as well, which we did, together with Visualway. We think they did a great job! Don't we look quite animally efficient now? (Which, by the way, is precisely what we strive for - animal efficiency - both in our doing and in the cababilities of our products.) 

We also moved all the information related to Otos, our people counting and utilization tracking service, to its dedicated site, www.otosservice.net. The reason for this is that the target audience for Otos Service is currently mainly Finnish municipalities, which is quite different compared to our other products and services. And that's why most of the content on the Otos site is currently available only in Finnish.

Last, but by no means least, we have launched a new product, Kide, an animally efficient quality supervisor for smaller sheet and web manufacturers. There's a lot more to tell you about that and we'll get back to the subject a little later.

That's all for now. Look for more soon!

(Posted by Heikki) 

This summer Intopii is into Beach Volley. Our Otos service counts spectators during the Beach Volley Pro tour from June 15 to August 25. The tour starts from Tampere and ends with the Finnish Championships at Helsinki. More information on the tour at biitsi.fi.

Since the Finnish parliamentary elections are drawing near, Intopii has decided to give Finnish voters an easy and entertaining way to find a suitable candidate. Naama Candidate Finder is based on Intopii's Naama Search Engine and it takes Candidate Finders into a totally new field. 

Please visit www.naama.fi to access Candidate Finder demo and additional information.

Oulu, Finland - Intopii Oy, a research and development firm specializing in intelligent technologies, located in Oulu, Finland, announced today that it has entered into a cooperative agreement with Cognex Corporation (NASDAQ: CGNX), the world's leading supplier of machine vision systems. Under the agreement, Intopii will work with Cognex to develop advanced analysis and classification tools for Cognex’s SmartView® surface inspection system.

"Cognex provides the leading surface inspection solution available today in paper, metals, nonwovens and plastics,“ said Markku Jaaskelainen, Senior Vice President and General Manager of Cognex's Surface inspection Systems Division. "The addition of Intopii’s advanced surface texture analysis methods to our high performance systems will provide additional benefits for our existing customers, while also enabling us to expand the reach of SmartView into new applications."

“We are very pleased to enter into cooperation with Cognex. This great opportunity is perfectly in line with our strategy to partner with industry leaders to provide added value to both OEMs and end users through our expertise and advanced technology”, said Janne Hetemaa, CEO of Intopii.

About Cognex

Cognex Corporation designs, develops, manufactures, and markets machine vision systems, or computers that can "see." Cognex is the world's leader in the machine vision industry, having shipped more than 75,000 machine vision systems, representing over $1.9 billion in cumulative revenue, since the company's founding in 1981. Cognex's Modular Vision Systems Division, headquartered in Natick, Massachusetts, specializes in machine vision systems that are used for automating the manufacture of a wide range of discrete items and for assuring their quality. Cognex's Surface Inspection Systems Division, headquartered in Alameda, California, specializes in machine vision systems that are used for inspecting the surfaces of products manufactured in a continuous fashion, such as metals, papers, and plastics. In addition to its corporate headquarters in Natick, Massachusetts, Cognex also has regional offices and distributors located throughout North America, Japan, Europe, and Southeast Asia. Visit Cognex on -line at http://www.cognex.com.

About Intopii

Intopii develops and offers products and services based on its extensive expertise in advanced intelligent technologies such as neural networks, pattern recognition, and computer vision. Intopii works closely together with end users, integrators, and OEMs on various industrial fields helping to find solutions to demanding challenges in quality assessment and process monitoring. Intopii’s advanced visual inspection software platform, Pinta, significantly facilitates addressing these challenges. Visit www.intopii.fi for further information.