Customer Projects
Pupillometer on a Smartphone
Pupillometry has much potential in medical care and prevention. To our knowledge, we have trained the first neural model for reliable detection of the pupill's centers and diameters. The data set has been labeled automatically and precisely with videos captured by a specifically design multi-camera system.
Ansprechperson: Prof. Dr. Martin Weisenhorn
Learning from few data
Multiple alternatives exist to capturing and labeling large sets of data. In this Project we have designed an object classifier based on classic and ML-based image processing techniques to classify with sufficient reliability for product rollout. Since then, large amounts of images, are captured automatically from the field, supporting the continuous fine tuning of classification networks.
Project partner: Tanzer Maschinenbau - Italy
Ansprechperson: Prof. Dr. Martin Weisenhorn