human-level visual physical reasoning
towards human-level interpretable visual physical reasoning and physical concept learning abilities on softbody and fluid under visual input
December 2022-September 2023
_________________________________________________________________
welcome to the project page here
_________________________________________________________________
the main challenge in this project was the definition of physical reasoning. What is the physical reasoning ability rooted in human mind? We've got the feel of reasoning basic rigid geometrical shapes in former works like [1] and [2]. But what about the softbody (rope, cloth, 3D softbody) and fluid? What is the definition of physical reasoning for them?
for most of time, I was seeking for an ideal explanation for the paradigm of human-level physical reasoning. the problem sounded very philosophical, but in fact it involved with technical limit.
sometimes the definition was not feasible for a procedural generation process. sometimes the definition was super hard for humans or super easy for machines. sometimes the simulation was too computationally demanding. ...references
1. Kexin Yi*, Chuang Gan*, Yunzhu Li, Pushmeet Kohli, Jiajun Wu, Antonio Torralba, and Joshua B. Tenenbaum. 2020. CLEVRER: CoLlision Events for Video REpresentation and Reasoning. In ICLR.
2. Daniel M. Bear*, Elias Wang*, Damian Mrowca*, Felix Binder*, Hsiao-Yu Fish Tung, R.T. Pramod, Cameron Holdaway, Sirui Tao, Kevin Smith, Fan-Yun Sun, Li Fei-Fei, Nancy Kanwisher, Joshua B. Tenenbaum, Daniel L.K. Yamins**, and Judith Fan**. 2021. Physion: Evaluating Physical Prediction from Vision in Humans and Machines. In NeurIPS.