Pégard Lab Student Wins 2020 Duke Research Computing Poster Competition
Mathew Atisa (poster first author and presenter, Class of 2023 Computer Science major, Pégard lab member) won first place in the 2020 Duke Research Computing Competition for their poster: Deep Learning with Specialized Training for Fast Computer Generated Holography.
The lab recently introduced a new algorithm for high speed computer generated holography that relies on deep learning to synthesize custom illumination patterns at record speeds. In this project, we compared the performance of our algorithm after training and testing our model on image datasets with custom shapes, e.g. squares and disks. Our results indicate that specialized learning with refined datasets significantly improves the quality of rendered images of the same subtype. Our findings benefit applications in biology and neuroscience where cell shape and size are known, and were computing holograms fast and accurately is critical.
Additional Poster Contributors: