Abstract:Falling is ranked highly among the threats in elderly healthcare, which promotes the development of automatic fall detection systems with extensive concern. With the fast development of the Internet of Things (IoT) and Artificial Intelligence (AI), camera vision-based solutions have drawn much attention for single-frame prediction and video understanding on fall detection in the elderly by using Convolutional Neural Network (CNN) and 3D-CNN, respectively. However, these methods hardly supervise the intermediate features with good accurate and efficient performance on edge devices, which makes the system difficult to be applied in practice. This work introduces a fast and lightweight video fall detection network based on a spatio-temporal joint-point model to overcome these hurdles. Instead of detecting fall motion by the traditional CNNs, we propose a Long Short-Term Memory (LSTM) model based on time-series joint-point features extracted from a pose extractor . We also introduce the increasingly mature RGB-D camera and propose 3D pose estimation network to further improve the accuracy of the system. We propose to apply tensor train decomposition on the model to reduce storage and computational consumption so the deployment on edge devices can to realized. Experiments are conducted to verify the proposed framework. For fall detection task, the proposed video fall detection framework achieves a high sensitivity of 98.46% on Multiple Cameras Fall, 100% on UR Fall, and 98.01% on NTU RGB-D 120. For pose estimation task, our 2D model attains 73.3 mAP in the COCO keypoint challenge, which outperforms the OpenPose by 8%. Our 3D model attains 78.6% mAP on NTU RGB-D dataset with 3.6× faster speed than OpenPose.

Human fall detection in surveillance video based on PCANet

Automatic Fall Detection of Human in Video Using Combination of Features

Video-based Fall Detection for Seniors with Human Pose Estimation

Learning Spatiotemporal Representations for Human Fall Detection in Surveillance Video.

Fall Detection for Elderly Person Care in a Vision-Based Home Surveillance Environment Using a Monocular Camera

Human Fall Detection Using Slow Feature Analysis

Video-Based Fall Detection Using Human Pose and Constrained Generative Adversarial Network

Visual Surveillance for Human Fall Detection in Healthcare IoT

Fall Detection Method based on a Human Electrostatic Field and VMD-ECANet Architecture

Elderly Falling Detection Based on Image Semantic Segmentation and CNN Model

Video Based Fall Detection Using Human Poses

A Computer Vision-Based Dual Network Approach for Indoor Fall Detection

A Fall Detection Network by 2D/3D Spatio-temporal Joint Models with Tensor Compression on Edge

Real-time video surveillance based human fall detection system using hybrid haar cascade classifier

Fall Detection Method for Infrared Videos Based on Spatial-Temporal Graph Convolutional Network

Fall Detection in Multi-Camera Surveillance Videos

Future Frame Prediction Network for Human Fall Detection in Surveillance Videos

Support Vector Machine Approach to Fall Recognition Based on Simplified Expression of Human Skeleton Action and Fast Detection of Start Key Frame Using Torso Angle.

A Novel Multi-Cue Integration System for Efficient Human Fall Detection

A human fall detection framework based on multi-camera fusion

Multi-camera, multi-person, and real-time fall detection using long short term memory