English

LA-Pose: Latent Action Pretraining Meets Pose Estimation

Computer Vision and Pattern Recognition 2026-05-01 v1

Abstract

This paper revisits camera pose estimation through the lens of self-supervised pretraining, focusing on inverse-dynamics pretraining as a scalable alternative to the current trend of fully supervised training with 3D annotations. Concretely, we employ inverse- and forward-dynamics models to learn latent action representations, similar to Genie from large-scale driving videos. Our idea is simple yet effective. Existing methods use latent actions in their original capacity, that is, as action conditioning of world-models or as proxies of robot action parameters in policy networks. Our method, dubbed LA-Pose, repurposes the latent action features as inputs to a camera pose estimator, finetuned on a limited set of high-quality 3D annotations. This formulation enables accurate and generalizable pose prediction while maintaining feed-forward efficiency. Extensive experiments on driving benchmarks show that LA-Pose achieves competitive and even superior performance to state-of-the-art methods while using orders of magnitude less labeled data. Concretely, on the Waymo and PandaSet benchmarks, LA-Pose achieves over 10% higher pose accuracy than recent feed-forward methods. To our knowledge, this work is the first to demonstrate the power of inverse-dynamics self-supervised learning for pose estimation.

Keywords

Cite

@article{arxiv.2604.27448,
  title  = {LA-Pose: Latent Action Pretraining Meets Pose Estimation},
  author = {Zhengqing Wang and Saurabh Nair and Prajwal Chidananda and Pujith Kachana and Samuel Li and Matthew Brown and Yasutaka Furukawa},
  journal= {arXiv preprint arXiv:2604.27448},
  year   = {2026}
}

Comments

Project page: https://la-pose.github.io/

R2 v1 2026-07-01T12:42:56.168Z