Estimation of gait normality index based on point clouds through deep auto-encoder

Abstract

This paper proposes a method estimating an index that indicates human gait normality based on a sequence of 3D point clouds representing the walking motion of a subject. A cylinder-based histogram is extracted from each cloud to reduce the number of data dimensions as well as highlight gait-related characteristics. A model of deep neural network is finally formed from such histograms of normal gait patterns to provide gait normality indices supporting gait assessment tasks. The ability of our approach is demonstrated using a dataset of 9 different gait types performed by 9 subjects and two other datasets converted from mocap data. The experimental results are also compared with other related methods that process different input data types including silhouette, depth map, and skeleton as well as state-of-the-art deep learning approaches working on point cloud.

Multi‐feature gait analysis approach using deep learning in constraint‐free environment

“…We can deduce the following from these findings. Overall, the suggested system's accuracy is equivalent to earlier studies in the sector (Li et al, 2018;Nguyen et al, 2016;Paiement et al, 2014).In (Nguyen et al, 2016;Nguyen & Meunier, 2018) authors deployed binary normal/pathological gait classification, they achieved 81% accuracy utilizing their database gathered by the Kinect device and 86% using combined Vicon/3D sensor data. In terms of gait assessment, the proposed characteristics outperform the skeletal joint locations.…”

Kinematic Covariance Based Abnormal Gait Detection

“…The [7] and [10] datasets perfectly suit the proposed algorithm specifics, since actors imitate different anomalies affecting the similarity of the gait. However, the number of sequences is small.…”

“…We enlarged the normal dataset with data coming from a gait recognition dataset, assuming that the people do not have a pathological gait, and by adding the Sphere dataset and the data acquired in our lab. We provide the results obtained on the [7] and [10] datasets separately and also perform some experiments on a mixed dataset as it was previously done by [5]. Information about the employed datasets is grouped in Table I.…”

“…Many researchers report the inaccuracy of feet position estimations and corresponding angles [9] coming from Kinect v.2. Available gait datasets [10], [11] also suffer from this issue. Li et al [8] proposed a method to classify normal and abnormal gait based on skeleton data.…”

A Combined Low-Rank Matrix Completion and CDBN-LSTM Based Classification Model for Lower Limb Motion Function

“…Nguyen T. N. and Meunier J. [12] collect point clouds of gait obtained from depth sensors, and propose Gait Normality Index (GNI), which is based on deep autoencoder, to indicate human gait normality. In [13] a twobranch multi-stage Convolutional Neural Network (CNN) is trained by Skeleton Gait Energy Image (SGEI) to accomplish the gait recognition task.…”