Synthetic data for understanding reality
Doctoral student at Linköping University, Sweden, with a focus on image synthesis and efficient data representations for computer graphics and visual machine learning. Research on visual scene understanding, photo-realistic image synthesis for automotive applications, synthetic datasets for material properties estimation and data augmentation strategies for boosting domain generalization in deep learning tasks. Strong interest and previous work in material shaders development and geometry reconstruction.
In the AI and machine learning era, data has proven to be both the constraining and the driving factor for effective applications. In a wide range of deep learning tasks, from object and material recognition to scene understanding and semantic segmentation for self-driving cars, context-rich training data seem to be one key to success. Apostolia will talk about synthetic datasets and how computer graphics can serve visual machine learning.
Hometown: Norrköping, Sweden