A physics-informed falling detection framework that recasts falling as a stability-loss event in a coupled dynamical system, yielding a sub-50K-parameter network capable of real-time inference on resource-constrained edge devices.
Abstract
Falling detection is vital for elderly care and intelligent surveillance; however, prevailing vision-based approaches predominantly frame it as static pose classification or discrete temporal pattern matching, fundamentally overlooking the instability dynamics of the human support system. This paper proposes a physics-informed falling detection framework that recasts falling as a stability-loss event in a coupled dynamical system. We introduce a novel dual-LTC architecture comprising a Center-of-Mass (CoM) subsystem and a Base-of-Support (BoS) subsystem, both instantiated as Liquid Time-Constant (LTC) neural networks to continuously model inertial trajectory evolution and ground-contact adjustment through adaptive time constants, Physical interpretability of falling motion. A learnable coupling module emulates physical interaction between the two subsystems, while a Stability Manifold classifier operates in the joint latent space to detect boundary crossing via Lyapunov-inspired stability metrics. Complementary counterfactual trajectory projection and Time-to-Collision (TTC) estimation further enable irreversibility assessment and early warning. The architecture is designed to support a three-state prediction paradigm (Normal, Falling, Fallen); in this preliminary study, we validate the core stability discrimination capability on a two-class dataset (Normal vs. Falling), leaving the full three-state temporal transition to future work. Unlike conventional CNN--RNN pipelines, the proposed formulation encodes continuous-time mechanical inertia, yielding a sub-50K-parameter network capable of real-time inference on resource-constrained edge devices. Extensive experiments demonstrate competitive accuracy with superior physical interpretability, validating its efficacy for low-compute visual fall detection.
The findings show that CNN-based architectures dominate algorithm choice, edge devices dominate deployment platforms, and optimization remains central to real-time inference on constrained hardware.
Mahammad Nabizade, Réda Yahiaoui, Isabelle Lajoie et al.· Italian National Conference...· 0 citations
Fall detection in multi-person environments, such as nursing homes and rehabilitation centers, is essential for ensuring the safety of vulnerable populations. Despite advances in deep learning, current vision- and skeleton-based fall detection systems often exhibit false negatives and reduced reliability in real-world scenarios due to scene complexity. This study presents a system-level analysis of fall detection errors by comparing four approaches—two skeleton-based methods using ST-GCN and ProtoGCN, a rule-based method, and a VIRA-GCN-based 3D joint method—on 95 RGB video sequences captured under minimally constrained multi-person conditions. We define six error types: skeleton structural interference, localized joint recognition failure, temporal skeleton identity inconsistency, object-to-skeleton association failure, viewpoint-induced observation limitation, and action-level ambiguity with similar activities. Although most methods achieved high event-level recall, their false-positive and false-negative patterns differed. The rule-based approach showed the most balanced performance, under the present experimental conditions, whereas the ST-GCN-based skeleton approach was more sensitive to joint-level and tracking instability. ProtoGCN reduced false positives but increased false negatives, showing a more conservative decision pattern. The VIRA-GCN-based 3D joint approach provided spatial cues but did not eliminate upstream pose and tracking errors. These results highlight the need for skeleton–depth fusion, robust identity tracking, occlusion handling, and enhanced joint recognition in real-world multi-person fall detection.
Han-Ye-Ang Lee, In-Nea Wang, Junho Jeong· Applied Sciences· 0 citations
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 modeling. Local processing performs segmentation and keypoint extraction; variational recurrent prediction and sequence classification then detect falls from observed and forecasted motion. We evaluate the framework on the UR Fall Detection and Human Fall datasets using random, subject-disjoint, and occlusion-based splits. Under random splits, neither representation consistently dominates, suggesting that standard protocols may hide meaningful differences. Under subject-disjoint evaluation, supervised keypoints show a statistically significant advantage, but performance varies by subject: they perform better when anatomical landmarks are visible, whereas unsupervised keypoints are more robust to occlusion and partial visibility, though they produce more false positives for complex activities. Under occlusion-based evaluation, supervised keypoints miss nearly half of all falls, while unsupervised keypoints retain strong sensitivity and substantially outperform them. Their anatomical independence allows spatial anchors to adapt to visible body structure rather than fail on absent landmarks. The gap widens under bandwidth constraints, where supervised localization errors compound through the temporal model. These findings show that representation choice should reflect expected visual conditions and that unsupervised keypoints offer an advantage when body visibility is compromised.
Tasmiah Haque, Jacob Kosinski, S. Mohan et al.· arXiv.org· 0 citations
The aging population has elevated falls into a critical public health issue. While camera-based YOLO algorithms offer non-contact detection, standard YOLOv13 struggles with occlusion, similar postures, and high computational demands. To address this, we propose RDD-YOLO, a task-oriented architecture optimized for fall detection accuracy and efficiency. Rather than simply stacking existing modules, RDD-YOLO assigns RepViTBlock to backbone feature extraction, DySample to detail-preserving feature fusion, and DHead to multi-scale regression, so that each component plays a complementary role in the detection pipeline. Evaluated on refined URFD and MCF datasets, RDD-YOLO outperformed YOLOv13, RT-DETR, and Faster-RCNN. For the URFD and MCF datasets, the mAP reached 92.1% and 87.3%, respectively. Speed tests showed that on a laptop (in a WSL2 environment), inference speed increased from approximately 47 FPS to approximately 57 FPS. Through pruning and FP16 half-precision inference, the speed further increased to over 70 FPS, demonstrating the feasibility of this method in real-time, resource-constrained application scenarios.
Gait classification and anomaly detection are non-intrusive approaches that can support healthcare surveillance and clinical gait analysis by identifying abnormal walking patterns. Despite recent advancements, existing methods remain limited by model complexity, high computational requirements, and privacy concerns. This study proposes a unified framework that combines three complementary components: (i) transformer-based temporal modeling to capture both short-term and long-term gait dynamics, (ii) lightweight architectures for efficient deployment on edge devices, and (iii) federated learning (FL) for privacy-preserving distributed training without requiring raw data sharing. Experiments were conducted on a balanced subset of 60,000 images from the Gait Detection Processed dataset using a controlled federated learning environment designed as a proof-of-concept evaluation rather than a large-scale deployment setting. The dataset consisted of three categories: background/non-gait, normal gait, and abnormal gait. Vision Transformer (ViT), ConvLSTM, and MobileViT architectures were evaluated for three-class gait classification and anomaly detection under a federated learning setting. Among the evaluated models, MobileViT-Large achieved the highest performance with 97.2% accuracy, 96.8% precision, 97.5% recall, and 97.1% F1-score, although it required higher computational resources and showed greater overfitting tendencies. MobileViT-Small achieved the best balance between efficiency and performance with 94.0% accuracy, making it more suitable for edge deployment. SHAP-based analysis further showed that the models focused on meaningful gait regions, such as torso and limb movements. This proposed framework provides a comparative benchmark of recurrent and transformer-based architectures within a privacy-preserving framework for healthcare monitoring applications.
Mohammad F Alfridi· Scientific Reports· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
