Identifying and curating trends is vital to stay competitive in the fashion and design industry. But finding the right samples from numerous images that are uploaded on different platforms every day is a difficult undertaking.

By combining web crawling with computer vision techniques our solution is able to find and filter relevant images based on content, color and shape. This allows our client to increase their catalogue of reference designs and to adjust to trends more quickly.

Although cost planning is the most crucial part when deciding on the profitability of new construction projects, the process of estimating costs is often unstructured and error prone with multiple planners having different approaches and no consistent database. This results in estimates being provided with safety factors of up to 40%.

After analysing and structuring the provided data, we developed a self-learning platform that drastically improved the cost estimation by leveraging on the automatic analysis of all historical projects - resulting in cost uncertainties of less than 10%.

Track beds are an essential component of a functioning railroad infrastructure. For timely maintenance, the condition of the subsoil must be analyzed by complex manual evaluation of existing sensor data. The quality of the analysis furthermore heavily depends on the experience of the evaluating person.

Using several thousand kilometers of sensor data, we developed an algorithm to automatically classify the track bed condition. This algorithm evaluates the sensor data with the click of a button and provides a consistent quality of the evaluation.

Engineering companies need to process vast amounts of documents to derive required information. This repetitive and error-prone task reduces productivity and limits the ability to act quickly when it matters, for example when submitting proposals before the competition.

By combining rule-based and machine-learning-based algorithms, keywords, relations and entities can be automatically extracted from documents. This enables companies to reliably abstract key information from unstructured data in a standardized and robust way.