Unsupervised Keypoints for Real-Time Fall Detection: Comparative Analysis Under Real-world Conditions with Predictive Bandwidth Reduction
Quick Answer
The study presents a privacy-preserving fall detection framework using unsupervised keypoints, outperforming supervised methods under occlusion and bandwidth constraints.
Quick Take
Evaluated on UR Fall Detection and Human Fall datasets, unsupervised keypoints maintain high sensitivity, while supervised keypoints struggle with visibility issues, missing nearly half of falls in occluded scenarios.
Key Points
- Unsupervised keypoints provide robust fall detection under occlusion, outperforming supervised methods.
- Supervised keypoints show a significant advantage when anatomical landmarks are visible.
- Under bandwidth constraints, supervised methods compound localization errors, reducing effectiveness.
- Evaluation on UR Fall Detection and Human Fall datasets reveals critical performance differences.
- Framework adapts to visible body structure, improving detection accuracy in challenging conditions.
Paper Resources
Source Excerpt
Falls among older adults are a major safety challenge, but continuous monitoring is difficult to sustain. Video captures fall-related posture and motion, yet deployment is limited by privacy, computation, and bandwidth. Supervised pose estimation is anatomically interpretable but vulnerable to occlusion and partial body visibility. We propose a privacy-preserving framework that replaces RGB transmission with compact motion representations based on unsupervised keypoints and predictive temporal m
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.