Matthias Niessner: Modeling 3D Shapes by Reinforcement Learning!
We train agents that aim to mimic human modeling steps by self supervision; agents are rewarded by evaluating the the goodness of the final model :)
Paper: https://arxiv.org/abs/2003.12397 https://t.co/oPgaVaQRKr
9 replies, 601 likes
roadrunner01: Modeling 3D Shapes by Reinforcement Learning
video: https://youtu.be/w5e9g_lvbyE https://t.co/xvWiGjy6XL
0 replies, 42 likes
Daisuke Okanohara: RL agents learn policies to model 3D shapes as human modelers; first composing basic primitives, then editing the meshes of the primitives. To facilitate the training, they combine heuristic policy, IL and RL https://arxiv.org/abs/2003.12397 https://www.youtube.com/watch?v=w5e9g_lvbyE
0 replies, 12 likes
Andrey Kurenkov 🤖: Very cool!
I find the idea of modeling the world in terms of primitives super interesting, it seems like it could be useful for tons of stuff.
2 replies, 11 likes
A Wojcicki: 3D ML with primitives and RL! 👍
0 replies, 2 likes
Found on Mar 30 2020 at https://arxiv.org/pdf/2003.12397.pdf