Subscribe to our monthly newsletter
CopernNet : Point Cloud Segmentation using ActiveSampling Transformers
In the dynamic field of railway maintenance, accurate data is critical. From ensuring the health of vegetation surrounding the tracks to maintaining the integrity of rails and poles, precise 3D data i …
June 24, 2024
Kapernikov enables Umicore to increase plant capacity
Kapernikov helped Umicore to make its precious metal recycling processes more efficient. This enabled the organization to increase its plant capacity and at the same time improve the safety of its wor …
May 4, 2023
Telraam: counting traffic with AI on affordable Edge devices
Systematically collecting traffic data can help city planners to find solutions for rat-run traffic, air pollution, traffic safety issues, and more. Together with traffic sensor specialist Telraam, Ka …
January 31, 2023
Reduce embedded systems development time using ROS
Need to create an automated machine that contains multiple sensors and motors to perform specific tasks (Aka ‘a robot’)? ROS offers a solid foundation for message brokering and plugins for a lot of ha …
December 8, 2022
Best practices for model training and serving in machine learning – Part 2
In this article series on machine learning, we discuss best practises for training and debugging your neural networks.
November 10, 2022
Cloud independent solutions
Vendor lock-in can be a nasty beast. And cloud-computing didn’t exactly make it easier to avoid. By asking the right questions, you can avoid import pitfalls and make a better trade-off.
October 28, 2022
Book review: what we learned from The Data-Centric Revolution
This book will change your life. Or at least the way you think about data management and enterprise information systems. Here’s our review of the book ‘The Data-Centric Revolution’.
October 13, 2022
From experimentation to production: how MLOps helps to deliver high-quality machine learning applications
A successful machine learning deployment depends on a complex interaction between data, machine learning model and code. MLOps is a good way to streamline this interaction. Let’s dive into some of t …
September 29, 2022
Best practices for model training and serving in machine learning – Part 1
In this article series on machine learning, we discuss best practises for training your data model.
August 19, 2022
Want to use AI on your problem, but struggling to get a good dataset? Part 2.
In article 2 of this series on how to get a good dataset for your machine learning project, we give you the best practices to get the most out of your data.
July 6, 2022
Three reasons why you shouldn’t rush blindly into deep learning
If you have a hammer, everything starts to look like a nail. Now that we have deep learning, it seems that every problem needs to be solved with it. Nothing could be further from the truth. Here are 3 …
June 22, 2022
Bart Van den Bergh
Bart joined Kapernikov in December 2019, a few months before a global pandemic brought the world to a standstill. Since then, he has worked as a software engineer on a wide variety of machine lea …
June 22, 2022
A comparison of stream processing frameworks
What stream processing framework will enable you to get instant insight into data as it flows? We compared a few common ones.
June 9, 2022
Want to use AI on your problem, but struggling to get a good dataset? Part 1.
Want to use AI on your problem, but struggling to get a good dataset? This series covers how to tackle that.
May 25, 2022
Future-proof your AI investment
Take advantage of new AI hardware & software without having to redevelop your machine learning model every three months.
March 25, 2022
High framerate deep learning at a low cost? Yes we can!
Real-time deep learning at a low cost. Impossible? High frame rates require expensive hardware? Learn how we got a $15 #Kendryte #K210 to run #YOLOv3 at a whopping 30 frames per second.
March 9, 2022
Louis Regout
Today, Louis Regout is mostly working on projects relating to asset data, AI on point clouds for Infrabel and AI development for the Telraam traffic counting project. “What I love most about Ka …
March 9, 2022
Traditional machine learning algorithms for machine vision
Today, traditional machine learning algorithms are significantly overshadowed by deep learning. However, they are still well suited for many applications independently or as a support in complex …
February 23, 2022
Which machine learning algorithm does your computer vision project need?
The recent advances in machine learning have given computer vision algorithms the ability to perform tasks with ever increasing speed and accuracy. But with a growing number of machine learning techni …
February 9, 2022
How we keep track of our data experiments
Experimentation is an essential part of machine learning. But how can you keep track of which model was trained on which dataset? At Kapernikov, we like to use Data Version Control (DVC), a dedicated …
January 26, 2022
5 first steps to implement continuous integration into your workflow
We may have convinced you earlier of the benefits of continuous integration (CI). But implementing this methodology into your business and workflows is another story. Where do you start? Right here.
January 20, 2022
AI on the Edge: why you (don’t) need to run your machine vision project in the cloud
Machine vision systems use large amounts of data that need to be stored and analyzed to make a decision. To do this effectively, businesses often turn to the cloud. But the cloud is not always the be …
December 14, 2021
3 ways to improve your AI machine vision results
Machine vision is booming, but does it fulfill your needs? Want to improve the performance of your AI-driven machine vision system? Here are three ways to influence and improve your results.
November 30, 2021
Moustafa Ayoub
Solving real problems Moustafa joined the Kapernikov ranks in June 2021. At that time, he already had two years of machine learning experience under his belt, and before joining the job market …
November 22, 2021
3 machine vision technologies you should know about
Machine vision systems do not always use standard cameras to inspect objects or to automate perception tasks. Did you already know these non-conventional machine vision technologies?
October 28, 2021
What is machine vision?
How does machine vision work? A camera collects the visual information, like the eye does.That collected camera image is processed for further use.A computer algorithm then analyzes the image, …
October 20, 2021
How to use DVC to experiment with data and annotated datasets
DVC can already be useful even in the exploration phase: while you are trying out different models, you are probably also building a good training and validation dataset. You want to make th …
October 14, 2021
Bart Verhagen
Bart is one of Kapernikov’s senior computer vision engineers, whose main role is to tackle challenges all the way from architecture to implementation.
September 14, 2021
What is continuous integration and why is it important for your business?
What is continuous integration? Continuous integration is the whole process where every software modification is thoroughly and automatically checked against your predefined, minimal quality s …
September 6, 2021
Why simulation makes perfect sense in machine vision projects.
Ever wondered how the Mars Lander was tested? Indeed, the lack of a real-life testing environment is a typical problem of space missions, but we see it in many other industrial settings as well. …
September 3, 2021
The DevOps 8
Eat – sleep – rave – repeat? Although DevOps also relies on a continuously repeating process, the steps do slightly differ! We explain them for human beings.
August 17, 2021
DevOps & data protection: how to work agile with sensitive data
We’ve said it before: we absolutely love DevOps. For Kapernikov, it’s a way to achieve quicker results, to deploy changes to a production environment much smoother, and to enable continuous del …
July 8, 2021
A comparison of image annotation tools
Deep learning relies on great training data. We took the time to put several annotation tools to the test.
June 23, 2021
Why Kapernikov loves DevOps (and why you should too)
Shorter development cycles, faster & more reliable delivery? Discover why Kapernikov loves to work the DevOps way, and why you should too.
June 15, 2021
Local kubernetes development
Developing cloud-native applications for Kubernetes gives you portability & scalability. But how do you get those pods to run on your laptop during development? We’ll explain your options in this tuto …
May 10, 2021
Co-development / co-creation @ Kapernikov
At Kapernikov, we are running many projects in a mixed setting: we work in teams consisting of our customers’ employees, Kapernikov colleagues and, occasionally, even other subcontractors. We have com …
April 29, 2021
First responders rely on high-quality railway data, from Ostend to Liège
When confronted with rail incidents, emergency responders need to be as fast and efficient as they can be. Especially in the case of bigger incidents, it is critical for first responders to achie …
December 1, 2020
Hans Nickisch
When Hans started his job as data management consultant at Kapernikov in January 2017, at first sight it may not appear to have been the most obvious choice for either of the parties. Hans came f …
November 4, 2020
How to make preventive maintenance work when you lack failure data
Manufacturing and utilities companies today usually have no shortage of data. Thanks to the proliferation of sensors and the internet of things, there is an almost unstoppable flood of readings. …
November 4, 2020
Kapernikov’s user-friendly reporting tool wins Asset Performance 4.0 hackathon
Leading up to the Asset Performance 4.0 conference, BEMAS organized three different hackathons about recent topics in maintenance. Kapernikov joined one of the hackathons dedicated to Fluvius. …
October 14, 2020
Mathieu Monet
Mathieu joined Kapernikov in February 2020. After a few years of learning the ropes at his first position as BI consultant, where he started to use ETL tooling and ended up building and managing a com …
October 14, 2020
Ludovic Santos
You might expect that studies in Chemistry may primarily lead to working in a lab, but Ludovic had already moved far beyond that as he was doing his PhD in Quantum Chemistry at the Université Li …
October 14, 2020
Visualizing railway vegetation detection
It has been some time since we finished the vegetation detection algorithm for Infrabel. The algorithm has proven its reliability so far, in that it detects both the presence and the height of the veg …
September 1, 2020
Complex data migrations done well
When organizations migrate to a new software application, usually a lot of attention goes to system selection and configuration. What typically generates less enthusiasm is making sure that the data t …
August 11, 2020
Synapse vs Snowflake
We wanted to compare the price/performance ratio of two of the leading data warehouse solutions in the Azure cloud scene: Microsofts own, Azure Synapse Analytics (a rebranded Azure SQL DW aiming to in …
August 3, 2020
A comparison of data processing frameworks
Data processing can be challenging, as sometimes it requires different execution environments and playing with different types of data. Orchestrating this data processing can be complex due to de …
July 29, 2020
Making your asset insights visual: Asset Performance hackathons finished & submitted
The final touches For the Fluvius case, we added a small Flask application to view filled in surveys as a report. For the Aquafin case, we took the opportunity to try some different analys …
June 16, 2020
Improving asset data reliability one (or two) challenge(s) at a time
After getting a feeling for the Aquafin pump station data, we took a step back. Of course it is always fun to play with data and see what is in there, but knowing what to look for tends to make t …
May 27, 2020
Asset management digitalization challenge accepted
The first challenge, presented by Aquafin, is a case on the detection of failures in pump stations. Being specialists in data science and asset management, this is exactly our cup of tea. The …
May 20, 2020
Consultants in quarantine
With tools like Git, Slack and Trello, we collaborate on code, share nice geek finds, discuss and organize without effort. Therefore, I expected the transition to the current lockdown would be r …
April 29, 2020
Cable Detection in Lidar Data
The catenary and contact wires are an essential component of the railway system, since they bring electric power to the trains through the pantographs. These cables usually consist of conductive mater …
April 7, 2020
Ipywidgets with matplotlib
A simple use case could be adding some basic controls to a plot for interactive data exploration. On the other side of the spectrum, we can combine widgets together to build full-fledged graphica …
March 31, 2020
Supporting drone inspections from a to z
To ensure the effective operation of an organization’s assets, asset managers need to plan regular inspections of their equipment. Drone inspections have big advantages over traditional, time-c …
March 9, 2020
Increasing asset intimacy
Why? Data-driven insights don’t come for free. Even though new AI-inspired technologies (data science, data analytics, machine learning) hold big promises, they don’t provide any real insights if …
February 25, 2020
Energy network campaign
Kapernikov assists Infrabel’s internal electrical grid operator in building an accurate and consistent overview of its cable assets and a topology of its networks. The end result is a high-qual …
February 17, 2020
Digital transformation is people business
How do people deal with new technologies? What is the employees’ perspective on company changes? And how can management make sure that new processes are fully adopted by the workforce? Those …
December 16, 2019
Kapernikov uses open data kit to seamlessly collect railway equipment data
Kapernikov has built up a considerable track record when it comes to data collection. But despite the company’s vast expertise, every data collection project is unique and requires a dedicated …
October 29, 2019
Writing a high-quality data pipeline for master data with apache spark – Part 2
In a previous article we introduced a number of best practices for building data pipelines, without tying them to a specific technology. Let’s see how this applies to several different technolo …
October 14, 2019
Writing a high quality data pipeline for master data with apache spark – Part 3
In a previous article, we explored a number of best practices for building a data pipeline. We then followed up with an article detailing which technologies and/or frameworks can help us adhere t …
October 14, 2019
Writing a high quality data pipeline for master data with apache spark – Part 1
At Kapernikov, we frequently work with master data. In this context, “working” means cleansing, analyzing, migrating and integrating. In order to do this, we frequently need to set up transfo …
October 8, 2019
Frank Dekervel
Frank is one of the founders of Kapernikov. Together with his business partner, Rein Lemmens, Frank started a web services agency in 2004. But with the addition of a third partner, Jan-Frederik Léger …
October 4, 2019
Drone work is teamwork
Even students with a heart for hardware want to look at things from a different angle from time to time. That’s why electronics engineering student Youri Vassiliev recently took on an internshi …
September 17, 2019
How to run a succesful data collection campaign
Kapernikov has been running a data collection campaign for Infrabel, Belgium’s railway network operator, for more than three years. And a new data collection campaign is in the pipeline. This h …
September 3, 2019
Establishing the digital railway
Level crossing equipment, signalling, track circuits, switches, incident reporting and much more. All this information is needed to efficiently manage maintenance processes and to monitor the con …
August 26, 2019
Deep learning sees whether people wear their safety helmets
Deep learning might seem like a complex, abstract concept, but it has helped Kapernikov to solve very concrete challenges. The Kapernikov team recently developed a Proof of Concept (PoC), based o …
July 10, 2019
Kapernikov’s data governance team is growing
Infrabel maintains and modernizes the rail network and controls the trains from its signal cabins. In order to plan and organize its everyday work, Infrabel needs reliable data about all of its a …
June 18, 2019
The future is teal
Growing pains Kapernikov is growing fast – in every possible way. We had an awesome 2018, and by the looks of it, 2019 will be no less exciting. Our growth is not limited to numbers, by the …
May 14, 2019
The art of making data work
IT projects have traditionally been a matter of delivering value through functionality. IT has always been about tools that allow people to work more efficiently or that provide better customer e …
April 1, 2019
Gödel’s incompleteness theorem
Large organizations often use different versions of the same master data. This is not always due to a wrong way of working, but rather the result of different use cases. This practice is supporte …
April 1, 2019
Kapernikov finds a new home base in Brussels
When entering the Pepibru business center in the heart of Brussels, the contrast between the old industrial building and the hip crowd of start-up companies and creative agencies immediately stri …
April 1, 2019
Nils Steyaert
Turning data into value Nils started freelancing for Kapernikov in 2016 and has since worked for well-known utility companies, such as Elia, Infrabel and Fluxys. Next to master data management …
April 1, 2019
Karel Van Camp
“You cannot know all the water in the sea, but you can learn how to swim.” It’s an attitude of life that has helped Karel Van Camp land a job at Kapernikov. Working as a data analyst for Kaperni …
April 1, 2019
Ubuntu on XPS15-9570
This manual will walk you through the installation of Ubuntu with a working CUDA setup on a Dell XPS15. We start from a Dell shipped with Windows and we use a Dual boot system.
March 26, 2019
Where consultants thrive
Just like many other IT consulting companies, Kapernikov offers consultants the opportunity to work for various interesting customers. However, we like to think that working at Kapernikov is a li …
March 11, 2019
Visual Odometry with the zed stereo camera
This tutorial briefly describes the ZED Stereo Camera and the concept of Visual Odometry. It also provides a step-by-step guide for installing all required dependencies to get the camera and visual od …
February 25, 2019
Debunking the myths about deep learning
Deep learning: breakthrough or bubble? Flemish engineering community ie-net recently organized an event around that topic. More specifically, ie-net’s expert group on industrial vision and r …
February 20, 2019
The Ros Robot_localization package
The robot_localization package is a collection of non-linear state estimators for robots moving in 3D (or 2D) space. (package summary – documentation) Each of the state estimators can fuse an arbitr …
January 24, 2019
Edge computing for industrial sensors and control systems
During a six-week internship, Kapernikov asked Arne Vlietinck to investigate whether Azure IoT Edge could be used to deploy, configure and monitor industrial sensors and control systems.
January 17, 2019
Improving asset management of the Belgian electrical grid with a data-driven approach
Kapernikov helped Elia to obtain an accurate overview of the condition of all its network equipment and to anticipate any risk of failure. As data consultants, we applied several database analysi …
January 10, 2019
Basic Introduction to computer vision
Computer Vision is the subfield of artificial intelligence which tries to imitate the human vision capabilities. And by “human vision”, we do not merely mean the eyes or the ability to see imag …
December 2, 2018
Our new legion of talents is here
For some of our talents, there seems to be a straight path from school to Kapernikov. For others, coming to Kapernikov is a result of life’s twists and turns. What’s your story? Nick De Sm …
November 27, 2018
Where employees come first
Kapernikov encourages all staff members to achieve their full potential Depending on individual needs and available project opportunities, Kapernikov strives to facilitate the professional dev …
November 19, 2018
Using Luigi to power a reporting pipeline
Luigi is an open source framework that has been designed to help you build, monitor and troubleshoot your data transformation pipeline. From running SQL queries, dumping results into a database, …
October 30, 2018
Kapernikov and euromaintenance 4.0: a match made in heaven
Four conference days, four presentations and multiple new contacts: the Kapernikov team is looking back at a very successful Euromaintenance 4.0 event. Not only was this a good opportunity to mee …
October 15, 2018
Object localization with a single camera and object dimensions
This tutorial shows how to do object localization in an image for an object of known dimensions. We will be using Python 3 with SciPy and OpenCV 3.1 on an Ubuntu 16.04 desktop system.
October 7, 2018
Kapernikov wins the 2018 ArcelorMittal Challenge in vision AI
“We are thrilled to have won this competition,” says Maarten De Munck, partner at Kapernikov. “The hackathon was an ideal way to work with innovative technologies that are completely i …
September 25, 2018
We are ready for euromaintenance 4.0. Are you?
Presentation: Supporting strategic decisions with data mining for Belgian Transmission System Operator Elia Time: Monday 24 September 2018 at 9 amHost: Julien Deuse (Kapernikov) Kapernikov …
September 11, 2018
Free training: introduction to C++
Sign up for the course Why C++? It’s not that we don’t love Python. It’s still our weapon of choice for data analytics, visualization and prototyping. I wish my children were as intui …
September 6, 2018
Kapernikov helps Infrabel to detect vegetation overgrowth in lidar data
The overgrowth of plants and weeds on railway lines poses a safety problem. Not only can these plants obstruct the driver’s view, but they can also obscure certain signals, interfere with overh …
August 31, 2018
Come learn from us at euromaintenance 4.0
Euromaintenance is Europe’s most important conference focusing on the 4th industrial revolution, IoT and predictive analytics for maintenance and condition monitoring. Euromaintenance 4.0 offer …
August 13, 2018
Multi-label classification with Keras
The test set contains only a few images, but includes a black dress (there were no black dresses in the training set), which was classified as a black jeans. Adrian’s explanation for this is: …
July 13, 2018
Normal computation using the GPU
Most frameworks for processing point clouds, like the Point Cloud Library (PCL), provide very convenient functionality for this computation. However, once we start scaling our point cloud, t …
June 27, 2018
Maarten De Munck
It has been a long time since Maarten was playing with his MSX computer and Fischertechnik construction toys as a young boy. Back then, it was all about making cool constructions and learning how …
June 17, 2018
Computer vision for flexible plant logistics
Robot-human interaction Lean manufacturing requires efficient collaboration between robots and humans. The flow of goods in a production process may look stable and fixed on paper. However, …
June 4, 2018
Tutorial: image classification with scikit-learn
For ease of reading, we will place imports where they are first used, instead of collecting them at the start of the notebook. This to prevent having to scroll up and down to check how an import …
April 10, 2018
Ezechiël Syx
Getting through a first job experience can be tough. Naturally, not everyone has years of work experience. Luckily, for Ezechiël, Kapernikov considers motivation and learning attitude to be equally i …
March 21, 2018
Kapernikov tackles point cloud processing in mini-hackathon
In a spirit of continuous improvement and knowledge sharing, Kapernikov recently organized another successful mini-hackathon event. This time, the Kapernikov team and a few colleagues from other compa …
February 22, 2018
Kapernikov 2.0.18
On this team day, next to a reception and a photo shoot for our new website, we organized an Italian cooking workshop. During the preparation of two pasta dishes, our consultants got to know each …
February 8, 2018
Customized computer vision & machine learning
From automated quality inspection tasks to robots working together with people, Kapernikov can develop the algorithms and build a proof-of-concept. Computer vision helps machines to detect objects in …
January 15, 2018
Kapernikov uses the power of whiteboarding
Whiteboards can be used to write or sketch on during meeting discussions or even when you’re alone trying to structure your thoughts. Since whiteboarding relies on a visual way of thinking, it …
December 19, 2017
Knowledge is power – Kapernikov invests in training on electricity grids
“Kapernikov takes the training of its employees very seriously,” says Hans Nickisch, data management specialist at Kapernikov. “Even if the training content is not directly applicable in yo …
October 19, 2017
Kapernikov puts AI to the test at the first ArcelorMittal Belgium hackathon
The challenge for Kapernikov was to see whether its AI algorithms could do better than a human operator in predicting when a weld would break. This welding process is an important step in the so- …
September 21, 2017
Kapernikov steels itself for the first ArcelorMittal hackathon
“This hackathon is a great way to challenge ourselves as a team,” says Rein Lemmens, partner at Kapernikov. “Data management and artificial intelligence in industrial settings are fields of …
September 14, 2017
ETL & data cleaning consultant wanted
When it comes to managing databases, we are a tower of strength. We deploy data cleansing campaigns in order to bring the corporate databases up to date again. We help our clients build new appli …
February 22, 2017
Practical workshop asset data integrity, data profiling & self service BI.
Join us in this workshop Gaining insights in the presence of data quality issues Enterprise data used in asset management commonly comprises data from different sources. Factors such as var …
October 14, 2016
Kapernikov joins the flink community
Apache Flink is a powerful and easy to use open source system for data stream processing with a very active developer and user community, one of the most active big data projects in the Apac …
September 16, 2016
Join the Kapernikov team as a data administrator
We’re on to something special here at Kapernikov. A fantastic team of data specialists who are all excited about technology, innovation and engineeringHappy customers all over Belgium wh …
August 16, 2016