OmniFall: From Staged Through Synthetic to Wild, A Unified Multi-Domain Dataset for Robust Fall Detection

TL;DR

OmniFall introduces a comprehensive, multi-domain dataset for fall detection, combining staged, synthetic, and wild videos to improve model robustness and generalization across diverse real-world scenarios.

Contribution

The paper presents OmniFall, a unified benchmark with diverse datasets and standardized protocols, facilitating research on privacy-preserving and generalizable fall detection models.

Findings

Synthetic data can match or outperform staged footage in cross-domain transfer.

Carefully designed synthetic data reduces privacy risks and eases data collection.

The benchmark enables development of robust fall detection models for uncontrolled environments.

Abstract

Visual fall detection models trained on small, staged datasets have unclear real-world utility due to limited diversity and inconsistent evaluation protocols. We present OmniFall, a unified benchmark with 80 hours / 15k videos and dense frame-level annotations in a harmonized 16-class taxonomy, spanning three complementary domains: OF-Staged (eight public staged sets, standardized with cross-subject/view splits), OF-Synthetic (12k videos, 17 h; controlled diversity in age, body type, environment, camera), and OF-In-the-Wild (the first test-only benchmark curated from genuine accident videos). OmniFall supports both video classification and timeline segmentation, and its cross-domain protocol isolates staged/synthetic-to-wild generalization. Our results show that carefully designed synthetic data can match or surpass real staged footage on cross-domain transfer, while reducing privacy risk and easing data collection. By combining privacy-amenable synthetic/staged sources with a public, test-only wild target and releasing dense, standardized timelines, OmniFall provides a comprehensive benchmark for privacy-preserving fall detection and fall-related (pre/post-fall) segmentation, enabling robust detectors that generalize to uncontrolled environments. Project page: http://simplexsigil.github.io/omnifall/