Computer Vision and Machine Learning
With world-class expertise in and experience with industrial machine vision, autonomous vehicles, and more, we provide complete solutions in computer vision, machine learning, and robotics.
Starting from your problem statement and input data specification, we design and build the correct custom solution. Where possible, we integrate maintained algorithm implementations from industry standard open source libraries (e.g. OpenCV, PCL, Ceres), but when necessary, we can develop performant custom implementations of academic work.
For applications requiring machine learning models, we work with popular frameworks such as TensorFlow and PyTorch. Our specialty is then in taking the next step, from prototype to robust, efficient, and maintainable production code.
During the discovery and analysis phase, we analyze our clients’ actual needs, and determine the correct solution architecture from a comprehensive knowledge of state-of-the-art techniques. In particular, we opt for traditional algorithms where the engineering costs trade off favorably against the long-term costs of computationally intensive machine learning solutions.
- Our solutions offer the best of both worlds: algorithms reflective of the academic state of the art integrated in truly production-ready implementations.
- Our solutions are maintainable, with clean architecture and encapsulated components that can be upgraded when improved algorithms become available.
- We prefer traditional algorithms where possible, for reduced long-term costs and increased control in comparison to machine learning solutions.