WiFi-JEPA: Self-supervised Learning for WiFi-CSI 3D Human Pose Estimation

2026-07-13

AI Summary

This work addresses the significant performance degradation of existing WiFi-based 3D human pose estimation methods under environmental variations and their reliance on costly camera-based annotations. To overcome these limitations, the authors propose a self-supervised learning framework that learns native representations of WiFi channel state information (CSI) by predicting masked CSI latent embeddings through masked modeling. Key innovations include a CSI-specific tensor tokenization scheme, a link-aware masking strategy, an unlabeled simulation pipeline for CSI generation based on ray tracing, and explicit modeling of cross-antenna-link correlations. Evaluated on the Person-in-WiFi-3D dataset, the proposed method achieves state-of-the-art performance in both single- and multi-person 3D pose estimation tasks, demonstrating the effectiveness of simulation-based pretraining for improving downstream performance in real-world scenarios.

Abstract

WiFi Channel State Information (CSI) enables privacy-preserving human pose sensing in camera-denied environments, but existing WiFi-based pose estimators often fail under environment shifts and rely on costly camera-based annotation pipelines that limit scale. We propose WiFi-JEPA, a self-supervised framework that learns CSI-native representations by predicting masked latent embeddings instead of reconstructing raw CSI signals that may contain hardware-specific artifacts. WiFi-JEPA makes three contributions: (i) CSI-specific tokenization and link masking tailored to the CSI tensor over channel, time, and link (C,T,L); masking entire Tx-Rx antenna links forces the model to predict one spatial link view from others, capturing cross-link correlations informative of 3D spatial structure. (ii) A ray-tracing CSI simulation pipeline that generates diverse unlabeled CSI from randomized geometric primitives, providing scalable pre-training data without pose annotations. (iii) State-of-the-art results on Person-in-WiFi-3D: WiFi-JEPA outperforms prior WiFi-CSI baselines on both single- and multi-person 3D pose estimation under the same evaluation protocol. We also show that simulated CSI provides complementary pre-training signal to real CSI, and that four vision-native SSL objectives degrade performance below training from scratch, whereas WiFi-JEPA consistently improves downstream pose estimation.

Problem

• WiFi-CSI
• 3D human pose estimation
• environment shift
• camera-based annotation
• privacy-preserving sensing

Innovation

• self-supervised learning
• WiFi CSI
• 3D human pose estimation
• ray-tracing simulation
• link masking