Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Neural Material Adaptor for Visual Grounding of Intrinsic Dynamics

About

While humans effortlessly discern intrinsic dynamics and adapt to new scenarios, modern AI systems often struggle. Current methods for visual grounding of dynamics either use pure neural-network-based simulators (black box), which may violate physical laws, or traditional physical simulators (white box), which rely on expert-defined equations that may not fully capture actual dynamics. We propose the Neural Material Adaptor (NeuMA), which integrates existing physical laws with learned corrections, facilitating accurate learning of actual dynamics while maintaining the generalizability and interpretability of physical priors. Additionally, we propose Particle-GS, a particle-driven 3D Gaussian Splatting variant that bridges simulation and observed images, allowing back-propagate image gradients to optimize the simulator. Comprehensive experiments on various dynamics in terms of grounded particle accuracy, dynamic rendering quality, and generalization ability demonstrate that NeuMA can accurately capture intrinsic dynamics.

Junyi Cao, Shanyan Guan, Yanhao Ge, Wei Li, Xiaokang Yang, Chao Ma• 2024

Related benchmarks

TaskDatasetResultRank
Future state predictionSpring-GS (test)
CD0.16
28
Object Dynamics GroundingSynthetic Object Dynamics Grounding Dataset
BouncyBall Score3.34
6
Initial state generalizationGorilla Real-world
PSNR29.78
5
Initial state generalizationReal-world dataset Chick2
PSNR28.92
5
Initial state generalizationReal-world dataset Chick1
PSNR30.73
5
Initial state generalizationReal-world dataset Mandarin
PSNR31.85
5
Initial state generalizationReal-world dataset Peanut
PSNR30.7
5
Initial state generalizationReal-world dataset Rabbit
PSNR28.3
5
Initial state generalizationReal-world dataset RBball
PSNR30.74
5
Intrinsic Dynamics ConsistencySandFish synthetic (test)
Chamfer Distance1.07
4
Showing 10 of 17 rows

Other info

Code

Follow for update