RAM: Recover Any 3D Human Motion in-the-Wild

TL;DR

RAM introduces a comprehensive framework combining semantic tracking, temporal priors, and predictive modeling to robustly recover 3D human motion in challenging, real-world scenarios.

Contribution

It presents a novel integrated system with adaptive filtering, memory-augmented modules, and future pose prediction for improved in-the-wild 3D human motion reconstruction.

Findings

Outperforms previous methods in Zero-shot tracking stability.

Achieves higher 3D accuracy on PoseTrack and 3DPW benchmarks.

Demonstrates robustness under occlusions and dynamic interactions.

Abstract

RAM incorporates a motion-aware semantic tracker with adaptive Kalman filtering to achieve robust identity association under severe occlusions and dynamic interactions. A memory-augmented Temporal HMR module further enhances human motion reconstruction by injecting spatio-temporal priors for consistent and smooth motion estimation. Moreover, a lightweight Predictor module forecasts future poses to maintain reconstruction continuity, while a gated combiner adaptively fuses reconstructed and predicted features to ensure coherence and robustness. Experiments on in-the-wild multi-person benchmarks such as PoseTrack and 3DPW, demonstrate that RAM substantially outperforms previous state-of-the-art in both Zero-shot tracking stability and 3D accuracy, offering a generalizable paradigm for markerless 3D human motion capture in-the-wild.