Sensible Deep Learning for 3D Data

Justin Solomon

Machine learning from 3D shapes is fundamentally different from learning on images. Rather than shoehorning popular image-based deep networks to take in shape data, we will build deep learning for point clouds, CAD models, and triangulated surfaces from the ground up. In particular, we will design, implement, train, and test networks built from operations that are mathematically well-posed and practically-minded specifically for geometric data. The end result is a general method for learning from 3D data of varying modalities and geometries, with application to object detection and analysis.

[June-1-2018 - current]



  1. Y. Wang, A. Fathi, A. Kundu, D. Ross, C. Pantofaru, T. Funkhouser, and J. Solomon, “Pillar-based Object Detection for Autonomous Driving,” in ECCV 2020, 2020 [Online]. Available:
  2. Y. Wang, Y. Sun, Z. Liu, S. E. Sarma, M. M. Bronstein, and J. M. Solomon, “Dynamic Graph CNN for Learning on Point Clouds,” ACM Trans. Graph., vol. 38, no. 5, pp. 1–12, Nov. 2019, doi: 10.1145/3326362. [Online]. Available:
  3. Y. Wang and J. Solomon, “Deep Closest Point: Learning Representations for Point Cloud Registration,” in 2019 IEEE/CVF International Conference on Computer Vision (ICCV), Seoul, Korea (South), 2019, pp. 3522–3531, doi: 10.1109/ICCV.2019.00362 [Online]. Available:
  4. Y. Wang, V. Kim, M. M. Bronstein, and J. Solomon, “LEARNING GEOMETRIC OPERATORS ON MESHES,” in ICLR 2019 Workshop on Representation Learning on Graphs and Manifolds, 2019 [Online]. Available:
  5. Y. Wang and J. M. Solomon, “PRNet: Self-Supervised Learning for Partial-to-Partial Registration,” in Advances in Neural Information Processing Systems 32, H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché-Buc, E. Fox, and R. Garnett, Eds. Curran Associates, Inc., 2019, pp. 8814–8826 [Online]. Available: