Aaron Michaux, graduate from the department of Electrical and Computer Engineering at Purdue University.

Supervisors and Research

My supervisor Dr Zygmunt Pizlo is a professor of psychological sciences who has, among many other things, trail-blazed research in the role of mathematical invariants (esp. symmetries) in shape perception. My co-supervisor Dr Edward Delp is the Charles William Harrison distinguished professor of electrical and computer engineering. In my research, I am interested in applying a cognitive science approach to engineering questions in computer vision. In particular I am interested in unsupervised object detection and 3D object reconstruction from real images using mathematical invariants — priors — that are inspired by research in perceptual psychology.

Research Interests

  • Figure/ground organization
  • Unsupervised object localization
  • Video tracking
  • 3D reconstruction
  • Computer vision
  • Object detection
  • 3D symmetry

My research interests evolved just after graduating, as I was reviewing the 2018 computer chess match between AlphaZero and Stockfish. As a chess enthusiast, I had some appreciation of AlphaZero’s dynamic play. Somehow the MCTS algorithm had internalized the “rules” of chess strategy, and amazingly, in a way that is comprehensible to humans. This led me to the conclusion that some machine learning approach could learn the “rules” of 3D object reconstruction, including symmetry invariants. Alas life caught up with me before being able to pursue this research project. A PhD is not long enough to do justice to the vast complexity of Computer Vision.

Conferences & Publications

Michaux, A. (2017). Two-view geometry, symmetry, and object perception. (Doctoral dissertation). Purdue University, West Lafayette, USA.

Michaux, A., Kuma, V., Jayadevan, V., Delp, E., & Pizlo, Z. (2017). Binocular 3D object recovery using a symmetry prior. Symmetry.

Jayadevan, V., Michaux, A., Delp, E., Pizlo, Z. (2017). 3D shape recovery from real images using a symmetry prior. Electronic Imaging.

Michaux, A., Jayadevan, V., Delp, E., & Pizlo, Z. (2016). Figure-ground organization based on three-dimensional symmetry. Journal of Electronic Imaging.

Palmer, E. Michaux, A., & Pizlo, Z. (2016). Using virtual environments to evaluate assumptions of the humanvisual system. 2016 IEEE Virtual Reality.

Jayadevan, V., Michaux, A., Delp, E., & Pizlo, Z. (2016). 3-D shape recovery from a single camera image. 2016 MODVIS Workshop.

Wang, Y., Higgins, N., Uleman, J., Michaux, A., & Vipond, D. (2015). An interactive activation and competition model of person knowledge, suggested by proactive interference by traits spontaneously inferred from behaviours. British Journal of Social Psychology.

Michaux, A., & Pizlo, A. (2015). Two correspondence problems easier than one. In Proceedings of the 2015 MODVIS Workshop.

Kwon, T., Li, Y., Scheessele, M., Michaux, A., & Pizlo, Z. (2014). Spatially-global interpolation of closed curves. In 4th Annual Midwestern Cognitive Science Conference.

Cao, Y., Barrett, D., Barbu, A., Narayanaswamy, S., Yu, H., Michaux, A., Lin, Y., Dickinon, S., Siskind, J., & Wang, S. (2013). Recognize human activites from partially observed videos. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.

Barbu, A. Bridge, A., Burchill, Z., Coroian, D., Dickinson, S., Fidler, S., Michaux, A., Mussman, S., Narayanaswamy, S., Salvi, D., Schmidt, L., Shangguan, J., Siskind, J., Waggoner, J., Wang, S., Wei, J., Yin, Y., & Zhang, Z. (2012). Video in sentences out. arXiv preprint arXiv:1204.2742v1

Barbu, A., Michaux, A., Narayanaswamy, S., and Siskind, J. (2010). Simultaneous object detection, tracking, and event recognition. arXiv preprint arXiv:1204.2741.

Teaching

I won the 2015 Magoon Award for excellence in teaching. I loved teaching every one of my students, and especially love seeing how far they have come since their Advanced C Programming courses. If you are one of them: I hope you enjoyed my class!