How to integrate your own ML vision models with Blaser’s hardware and software?
How to monitor a production line, anonymize recordings, or easily browse through months of entries to find the one that refers to an investigated complaint? Basler’s clients have more and more use cases to cover, some of them applicable only to their own machine learning models.
Basler is leading the way in image processing and computer vision hardware and software, providing hundreds of companies worldwide with advanced camera suites. The company saw a business opportunity in adding Machine Vision to their portfolio, hence allowing their clients to use their own ML models and integrate them with Basler’s hardware and software.
Since our previous collaboration with Basler’s Head of R&D turned out successful, the company decided to reach out to us to address the challenges. Our team had vast experience in starting new and non-standard software projects (and then developing them on an ongoing basis).
What was the challenge?
If you fail to plan, you plan to fail. How we set the base for advanced software development.
The ability to translate ideas and vision into a tangible plan was what differentiated us from other companies Basler talked to.
In the course of our collaboration, after the release of the MVP, we continued the development, enhancing the application with new features and functionalities. The development of the application is ongoing, with Codya support being one of the elements of the process.
We composed an advanced team with software developers, DevOps, UX/UI designer, and Business Analyst. The process took less than a month, which was a few times faster than recruiting the team in-house.
To start on a high note, we arranged a series of regular workshops and meetings with Basler specialists to collect and specify both business and technical needs for the web application. That’s also when the first mockups in Figma and architecture concepts in Azure Cloud were born.
For the next several weeks, we created a scalable development process that allowed us not only to start the development but also to accommodate a growing team, which we extended a few months later.
With them in hand, we cherry-picked the technologies that would meet the project’s requirements (microservices and containers being some of them).
In the meantime, our team created interactive mockups of pylon AI that Basler could present at the Vision Show.
Then, we published the application, with regards to Basler’s monitoring and security requirements.
The client was satisfied with the results, therefore we moved to a stable and regular development of new features and functionalities, including calculating statistics and generating graphs, or implementing classification use cases.
We composed an advanced team with software developers, DevOps, UX/UI designer, and Business Analyst. The process took less than a month, which was a few times faster than recruiting the team in-house.
To start on a high note, we arranged a series of regular workshops and meetings with Basler specialists to collect and specify both business and technical needs for the web application. That’s also when the first mockups in Figma and architecture concepts in Azure Cloud were born.
With them in hand, we cherry-picked the technologies that would meet the project’s requirements (microservices and containers being some of them).
For the next several weeks, we created a scalable development process that allowed us not only to start the development but also to accommodate a growing team, which we extended a few months later.
In the meantime, our team created interactive mockups of pylon AI that Basler could present at the Vision Show.
Then, we published the application, with regards to Basler’s monitoring and security requirements.
The client was satisfied with the results, therefore we moved to a stable and regular development of new features and functionalities, including calculating statistics and generating graphs, or implementing classification use cases.
pylon AI makes it possible to verify the client’s model on different hardware platforms and finally choose the one with the best price-to-quality ratio.
Users can upload their models and datasets to choose proper hardware configurations. The application will generate ready-to-use installation bundles for chosen Basler hardware.
Microservice architecture we designed is more fault-tolerant and reduces the number of dependencies between particular parts of the application.
We used Azure Cloud to speed up the development process and reduce implementation. We used Azure services as storage, queue server, database, authentication, and more.
Basler customers can use the power of Machine Learning.
application screens
months of Design Phase
months of MVP development
months of support and regular application development
months of MVP development
months of Design Phase
application screens
months of support and regular application development
Technology stack
The enhancement of machine learning and computer vision is what sets this project apart. On top of that, a successful and long-term collaboration with Basler shows that Codya is capable of working hand in hand with large, international enterprises seeking to build their competitive advantage through technology.
Other case studies
You have a challenge. We’re there to solve it.
Let’s grab a (virtual) coffee and discuss your project.
