Wearable Camera- and Accelerometer-Based Fall Detection on Portable Devices

Abstract : Robust and reliable detection of falls is crucial especially for elderly activity monitoring systems. In this letter, we present a fall detection system using wearable devices, e.g., smartphones, and tablets, equipped with cameras and accelerometers. Since the portable device is worn by the subject, monitoring is not limited to confined areas, and extends to wherever the subject may travel, as opposed to static sensors installed in certain rooms. Moreover, a camera provides an abundance of information, and the results presented here show that fusing camera and accelerometer data not only increases the detection rate, but also decreases the number of false alarms compared to only accelerometer-based or only camera-based systems. We employ histograms of edge orientations together with the gradient local binary patterns for the camera-based part of fall detection. We compared the performance of the proposed method with that of using original histograms of oriented gradients (HOG) as well as a modified version of HOG. Experimental results show that the proposed method outperforms using original HOG and modified HOG, and provides lower false positive rates for the camera-based detection. Moreover, we have employed an accelerometer-based fall detection method, and fused these two sensor modalities to have a robust fall detection system. Experimental results and trials with actual Samsung Galaxy phones show that the proposed method, combining two different sensor modalities, provides much higher sensitivity, and a significant decrease in the number of false positives during daily activities, compared to accelerometer-only and camera-only methods.

 EXISTING SYSTEM :

 ? Most of the existing studies use falls simulated in a laboratory environment to test the obtained performance. ? As a different research area, the incorporation of fall detection systems into existing electronic devices (such as hearing aids, cochlear implants, or pacemakers) could be an unobtrusive way to detect falls. ? Most fall detection studies used simulated falls (usually performed by young subjects) to create and validate the proposed fall detection systems. ? In this way, in fact, when training with four of the five folds and testing with the remaining fold, it is assured that data from subjects of the testing fold is only in the testing fold, thus preventing possible overfitting.

 DISADVANTAGE :

 ? In such cases, use of camera sensors in tandem with the accelerometer can help resolving such issues. ? A simulated environment can indeed mask a series of possible problems that can disclose in real-world settings. ? In fact, performed in a controlled environment, which provides a standardized and clean version of the problem of fall detection. ? Such an approach leads to generalizability and transferability issues of the results from simulated to real environments, a topic covered in recent papers. ? In contrast, in an uncontrolled environment, several procedural and technical issues need to be faced.

 PROPOSED SYSTEM :

 • In this letter, we propose a system that employs a novel approach of using both accelerometer and camera modalities to detect falls by differentiating them from other daily activities including walking, sitting, lying down and going up and down stairs. • We first present the significant improvement provided by the proposed camera-based algorithm on recorded videos. • In the original HOG-based algorithm, proposed for human detection, an image is divided into blocks and then each block is divided into cells. • In, we proposed a modified HOG algorithm for fall detection, wherein, different from original HOG, separate histograms are constructed for edge strength (ES) and edge orientations (EO).

 ADVANTAGE :

 ? Our objective in this paper is to analyze the performance of state-of-the-art machine learning algorithms on newly proposed features when benchmarked on a relatively large database of real-world falls. ? We performed subject-based cross-validation to evaluate the performance of the classifiers. ? The performance of the classifiers was evaluated by computing several measures. ? When considering the obtained performance results, features from the multiphase model overperformed conventional features for each classifier, both in F-measure and AUC values. ? This highlights the importance of external validation of fall detection algorithms, and the importance of reporting comparable performance measures across studies.