Expertise

Search Technologies and Interactive Information Systems

We work with modern search technologies and develop customer-specific search engines of various kinds:

• Full-text search in millions of documents by using inverted search indices based on Open‑Source technologies
• Problem-specific ranking functions that improve the relevancy of search results
• Automatic extension of search queries with synonyms and related terms such that documents can be found, which contain specialized words not familiar to a user
• Similarity search of images, documents and products, e.g., as part of a product recommendation system
• Innovative extensions for explorative search, e.g., interactive search maps that provide an overview over the whole set of results as demonstrated by our product “Atlas”

We rely on a number of existing Open‑Source tools, including Apache Lucene, Apache Solr, PostgreSQL, TensorFlow, NLTK, ElasticSearch and Vespa.

AI & Machine Learning

Modern methods of artificial intelligence have great potential to revolutionize many business areas. We deal with many state‑of‑the‑art machine learning algorithms in different scenarios:

• Classification of documents, images, and products based on statistical models like Support Vector Machines and methods from the area of Deep Learning
• Recommender systems for products and documents by using content-based similarity analysis and search
• Time series forecasting of various reference numbers by using regression analysis

For implementing machine learning models, we rely on Open‑Source technologies like Tensorflow, PyTorch, Transformers, Gensim, SciPy, Pandas, Scikit-Learn, and many more.

Cloud & Container Management

To provide a smooth operation of our software modules, we rely on container and cloud technologies, for example Docker and Kubernetes, which offer numerous advantages in comparison to classic server-based operation.

• Splitting software into modules and independent containers benefits flexible and rapid software development cycles both of each module individually, and as a consequence, also regarding the whole software system.
• Container can be transferred easily between servers and providers, such that both server and maintenance expenses can be reduced.
• In the case of high work loads, individual components, e.g., a database or search index, can be easily deployed on multiple servers to distribute the workload.

We support you in both developing new cloud-based applications, but also in updating legacy systems. We are happy to discuss with you how popular cloud providers such as Amazon AWS, Google Cloud and Microsoft Azure can accelerate your development process in the context of your specific application scenario.

Open‑Source Technologies

Today, many software solutions are based to a great extent on Open‑Source technologies. On the one hand, application logic often uses advanced Open‑Source software libraries. On the other hand, the management and organization of the software development process uses Open‑Source tools.

We offer our extensive experience in many Open‑Source technologies:

• Open‑Source tools that allow to efficiently develop software, such as Git, GitLab, Visual Studio Code, Eclipse, and many more
• Open‑Source solutions that facilitate a smooth operation of applications, such as Linux, PostgreSQL, Docker, Kubernetes, and many more.
• Open‑Source programming libraries that provide advanced algorithms for scientific data analysis, image processing, document processing, visualization, and many more

We are happy to discuss with you how Open‑Source technologies can support you in accelerating your next software project.

Industrial Image Processing for Real‑Time Defect Detection

We also have significant experience in the area of automated image processing of recordings taken by industrial camera systems:

• Configuration, integration, and calibration of industrial cameras offered by companies like IDS, Baumer, Balluff, Basler supporting the GigE Vision and GenICam® standard
• Real-time detection of defects on reflective surfaces based on image segmentation implemented as GPU shaders
• Automated localization and visualization of recognized defects using virtual 3D models
• Classification of different types of defects like inclusions or scratches based on machine learning models
• Integration with industrial automation systems by communicating with Programmable Logic Controllers (PLC) over OPC-UA or ModBus

For image analysis and defect detection, we rely on many technologies, e.g., OpenGL, OpenCV, Scikit-Image, PyTorch, and many more.