UP-Fall Detection Dataset: A Multimodal Approach

Abstract

Falls, especially in elderly persons, are an important health problem worldwide. Reliable fall detection systems can mitigate negative consequences of falls. Among the important challenges and issues reported in literature is the difficulty of fair comparison between fall detection systems and machine learning techniques for detection. In this paper, we present UP-Fall Detection Dataset. The dataset comprises raw and feature sets retrieved from 17 healthy young individuals without any impairment that performed 11 activities and falls, with three attempts each. The dataset also summarizes more than 850 GB of information from wearable sensors, ambient sensors and vision devices. Two experimental use cases were shown. The aim of our dataset is to help human activity recognition and machine learning research communities to fairly compare their fall detection solutions. It also provides many experimental possibilities for the signal recognition, vision, and machine learning community.

Fall Detection Using Multiple Bioradars and Convolutional Neural Networks

“…Unlike other works in which falls were performed at the same place of the experimental scene [18,[22][23][24], in the present work we try to solve a more challenging task, namely to detect falls performed at different positions. To do so, the volunteer was asked to enter the room and imitate falling or normal daily activity while reaching one of four points (points 1-4 in Figure 4), thus the range between the subject and the radars varied from 1.0 to 2.0 m, moreover, for points 3 and 4 in Figure 4, the volunteer was only partially observed by bioradar No.…”

“…In most papers dealing with fall detection problem, it is proposed to select features from the raw data, construct a feature vector with a feature extraction technique, and use this vector for training a classifier [18,22,24]. Feature selection and extraction are methods used to convert the raw data into a low-dimensional subspace that contains all relevant information for a further classification step [33].…”

“…Some research groups proposed to used multi-sensor approaches which combine both contact and non-contact sensors to overcome the drawbacks of the previously mentioned methods [17,18]. However, such multi-sensors systems encounter more difficulties in implementing them in the field than systems utilizing only a single modality of sensor.…”

Internet of Things (IoT)-Enabled Elderly Fall Verification, Exploiting Temporal Inference Models in Smart Homes

“…Consequently, contemporary research in the field of fall detection and verification of the elderly and impaired population is extensive. Among the critical challenges and issues reported in the literature is the difficulty of an accurate comparison between fall detection systems and machine learning techniques for detection [12].…”

“…In response to a few multimodal datasets developed upon different human activities, including publicly available falls, Martinez-Villasenor et al presented the UP-Fall Detection Dataset. The authors decided to choose a 10-fold configuration based on relevant research and everyday practices reported in machine learning [12]. The dataset was adaptable enough to summarize full information from wearable sensors, ambient sensors, and vision devices.…”