Neural Implicit Representations for Physical Parameter Inference from a Single Video
This addresses the challenge of data efficiency and generalization in physics-based vision for applications like robotics and simulation, though it builds incrementally on existing neural representation and ODE methods.
The paper tackles the problem of inferring physical parameters from visual data by proposing a model that combines neural implicit representations with neural ODEs, enabling identification from a single video and achieving photo-realistic synthesis and interpretable parameter extraction.
Neural networks have recently been used to analyze diverse physical systems and to identify the underlying dynamics. While existing methods achieve impressive results, they are limited by their strong demand for training data and their weak generalization abilities to out-of-distribution data. To overcome these limitations, in this work we propose to combine neural implicit representations for appearance modeling with neural ordinary differential equations (ODEs) for modelling physical phenomena to obtain a dynamic scene representation that can be identified directly from visual observations. Our proposed model combines several unique advantages: (i) Contrary to existing approaches that require large training datasets, we are able to identify physical parameters from only a single video. (ii) The use of neural implicit representations enables the processing of high-resolution videos and the synthesis of photo-realistic images. (iii) The embedded neural ODE has a known parametric form that allows for the identification of interpretable physical parameters, and (iv) long-term prediction in state space. (v) Furthermore, the photo-realistic rendering of novel scenes with modified physical parameters becomes possible.