Pith. sign in

REVIEW 7 cited by

CoPhy: Counterfactual Learning of Physical Dynamics

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 1909.12000 v2 pith:2SKATHIW submitted 2019-09-26 cs.CV

CoPhy: Counterfactual Learning of Physical Dynamics

classification cs.CV
keywords physicalcounterfactuallearningalternativecophydynamicsenvironmentlatent
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Understanding causes and effects in mechanical systems is an essential component of reasoning in the physical world. This work poses a new problem of counterfactual learning of object mechanics from visual input. We develop the CoPhy benchmark to assess the capacity of the state-of-the-art models for causal physical reasoning in a synthetic 3D environment and propose a model for learning the physical dynamics in a counterfactual setting. Having observed a mechanical experiment that involves, for example, a falling tower of blocks, a set of bouncing balls or colliding objects, we learn to predict how its outcome is affected by an arbitrary intervention on its initial conditions, such as displacing one of the objects in the scene. The alternative future is predicted given the altered past and a latent representation of the confounders learned by the model in an end-to-end fashion with no supervision. We compare against feedforward video prediction baselines and show how observing alternative experiences allows the network to capture latent physical properties of the environment, which results in significantly more accurate predictions at the level of super human performance.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Do generative video models understand physical principles?

    cs.CV 2025-01 unverdicted novelty 8.0

    Physics-IQ benchmark reveals that generative video models exhibit limited physical understanding unrelated to their visual quality.

  2. YoCausal: How Far is Video Generation from World Model? A Causality Perspective

    cs.CV 2026-05 unverdicted novelty 7.0

    YoCausal benchmark shows video diffusion models detect the arrow of time but lack genuine causal understanding relative to humans.

  3. $\Delta$ynamics: Language-Based Representation for Inferring Rigid-Body Dynamics From Videos

    cs.CV 2026-05 unverdicted novelty 6.0

    A vision-language framework generates text-based rigid-body scene configurations from videos using motion reasoning and optical flow, reporting 0.30 IoU on CLEVRER (7x over baselines) and transfer to 235 real videos.

  4. Ada-Diffuser: Latent-Aware Adaptive Diffusion for Decision-Making

    cs.LG 2026-05 unverdicted novelty 6.0

    Ada-Diffuser is a causal diffusion model that jointly learns observed interaction structure and underlying latent dynamics from minimal observations for adaptive planning and policy learning.

  5. PhyCo: Learning Controllable Physical Priors for Generative Motion

    cs.CV 2026-04 unverdicted novelty 6.0

    PhyCo adds continuous physical control to video diffusion models via physics-supervised fine-tuning on a large simulation dataset and VLM-guided rewards, yielding measurable gains in physical realism on the Physics-IQ...

  6. IMBench: A Benchmark for Intuitive Robotic Manipulation

    cs.RO 2026-07 conditional novelty 5.0

    IMBench is a 35-task robosuite benchmark with a three-stage evaluation showing current VLMs and robot policies fail to convert physical reasoning into executable manipulation.

  7. Physics-IQ Verified

    cs.CV 2026-06 unverdicted novelty 5.0

    Physics-IQ Verified refines 57.6% of samples and 34.8% of prompts from the original benchmark and produces moderate ranking shifts (Kendall's τ = 0.46) across six image-to-video models.