PARADOX studios - principal AI researcher
My role in this project as principal AI researcher includes developing an attribute conditioned instance retrieval pipeline that generates image sequences based on attribute corelation intensity. The show is about Frank van der Salm’s work from the past three decades, which is a statement about an urbanized world that seems to be modelled after its very own images in photography, video, movies, games, news and entertainment. It results in a strong dystopian feeling; factual, sociological or topographical references are rare. In Van der Salm’s universe, buildings and urban areas are part of one global, imaginary city. The sequences generated from my model will be projected on a 25mx3m wall in the Rotterdam Photo Musem starting January 2020. See you there!
NICAS: Netherlands Institute for Conservation, Art and Science
This project works on visual analytic methods aimed at creating synergy between humans and machines in performing data-intensive problems. Often such methods are developed by computer scientists and evaluated with students within computer science or by using crowdsourcing involving participants which are not in computer science, so without expertise in the relevant domain. The VISTORY project allows to develop methods and to evaluate the computer generated result in close conjunction with art history experts. It creates a unique opportunity to develop methodologies to enhance pattern recognition in digital image collections by truly integrating the analytic capabilities of humans and machines, a long standing goal of visual analytics research.
Developing expert systems for image classification and retrieval in the medical and biological domain.
- PlantBot - A plant species classification engine
- Multimodal medical image retrieval
- Hand gesture recognition from a live video source
- EM brain tissue image segmentation