Gait Segmentation Research Articles

Adaptive Detection in Real-Time Gait Analysis through the Dynamic Gait Event Identifier.

The Dynamic Gait Event Identifier (DGEI) introduces a pioneering approach for real-time gait event detection that seamlessly aligns with the needs of embedded system design and optimization. DGEI creates a new standard for gait analysis by combining software and hardware co-design with real-time data analysis, using a combination of first-order difference functions and sliding window techniques. The method is specifically designed to accurately separate and analyze key gait events such as heel strike (HS), toe-off (TO), walking start (WS), and walking pause (WP) from a continuous stream of inertial measurement unit (IMU) signals. The core innovation of DGEI is the application of its dynamic feature extraction strategies, including first-order differential integration with positive/negative windows, weighted sleep time analysis, and adaptive thresholding, which together improve its accuracy in gait segmentation. The experimental results show that the accuracy rate of HS event detection is 97.82%, and the accuracy rate of TO event detection is 99.03%, which is suitable for embedded systems. Validation on a comprehensive dataset of 1550 gait instances shows that DGEI achieves near-perfect alignment with human annotations, with a difference of less than one frame in pulse onset times in 99.2% of the cases.

Real-time Gait Event Tracker (RGET): An Innovative Approach to Digital Human Modeling through Integration of Inertial Measurement Unit and Dynamic Feature Extraction

The advancement of digital human modeling technology underscores the importance of accurately representing human characteristics and behaviors using motion capture technologies. This paper introduces the Real-time Gait Event Tracker (RGET), an advanced method aimed at precisely detecting gait events via an Inertial Measurement Unit (IMU), thereby supporting high-precision digital human modeling. RGET integrates planning techniques into continuous states and action spaces, utilizing first-order difference functions and sliding window techniques to effectively identify four key gait events: Heel Strike (HS), Toe Off (TO), Walking Start (WS), and Walking Pause (WP). The system efficiently manages IMU gait signal data through real-time queuing techniques and extracts dynamic features within time frames using positive/negative windowed first-order difference integrations. By adopting weighted sleep time methods and adaptive threshold decision rules, RGET accurately extracts gait event features from sequences. Not only does RGET support precise gait segmentation with an F1 score of 95.9%, but it also provides an innovative method for digital human modeling and its applications.

Deep Learning-Assisted Gait Parameter Assessment for Neurodegenerative Diseases: Model Development and Validation.

Neurodegenerative diseases (NDDs) are prevalent among older adults worldwide. Early diagnosis of NDD is challenging yet crucial. Gait status has been identified as an indicator of early-stage NDD changes and can play a significant role in diagnosis, treatment, and rehabilitation. Historically, gait assessment has relied on intricate but imprecise scales by trained professionals or required patients to wear additional equipment, causing discomfort. Advancements in artificial intelligence may completely transform this and offer a novel approach to gait evaluation. This study aimed to use cutting-edge machine learning techniques to offer patients a noninvasive, entirely contactless gait assessment and provide health care professionals with precise gait assessment results covering all common gait-related parameters to assist in diagnosis and rehabilitation planning. Data collection involved motion data from 41 different participants aged 25 to 85 (mean 57.51, SD 12.93) years captured in motion sequences using the Azure Kinect (Microsoft Corp; a 3D camera with a 30-Hz sampling frequency). Support vector machine (SVM) and bidirectional long short-term memory (Bi-LSTM) classifiers trained using spatiotemporal features extracted from raw data were used to identify gait types in each walking frame. Gait semantics could then be obtained from the frame labels, and all the gait parameters could be calculated accordingly. For optimal generalization performance of the model, the classifiers were trained using a 10-fold cross-validation strategy. The proposed algorithm was also compared with the previous best heuristic method. Qualitative and quantitative feedback from medical staff and patients in actual medical scenarios was extensively collected for usability analysis. The evaluations comprised 3 aspects. Regarding the classification results from the 2 classifiers, Bi-LSTM achieved an average precision, recall, and F1-score of 90.54%, 90.41%, and 90.38%, respectively, whereas these metrics were 86.99%, 86.62%, and 86.67%, respectively, for SVM. Moreover, the Bi-LSTM-based method attained 93.2% accuracy in gait segmentation evaluation (tolerance set to 2), whereas that of the SVM-based method achieved only 77.5% accuracy. For the final gait parameter calculation result, the average error rate of the heuristic method, SVM, and Bi-LSTM was 20.91% (SD 24.69%), 5.85% (SD 5.45%), and 3.17% (SD 2.75%), respectively. This study demonstrated that the Bi-LSTM-based approach can effectively support accurate gait parameter assessment, assisting medical professionals in making early diagnoses and reasonable rehabilitation plans for patients with NDD.

Quantitative analysis of interface pressures in transfemoral prosthetic sockets.

Among the different factors affecting socket comfort, the pressure applied on residual limb tissues is a crucial parameter for the success or failure of any prosthetic device. However, only a few incomplete data are available on people with transfemoral amputation, in this regard. This work aims at filling this gap in the literature. Ten people with transfemoral amputation wearing 3 different socket designs were recruited in this study: 2 ischial containment sockets featured by proximal trim lines that contain the ischial tuberosity and ramus and greater trochanter, 2 subischial sockets with proximal trim lines under the ischium level, and 6 quadrilateral sockets with proximal trim lines that contain the greater trochanter and create a horizontal seat for the ischial tuberosity. The pressure values at the anterior, lateral, posterior, and medial areas of the socket interface were recorded during 5 locomotion tasks (ie, horizontal, ascent, and descent walking, upstairs and downstairs) by using an F-Socket System (Tekscan Inc., Boston, MA). Gait segmentation was performed by exploiting plantar pressure, which was acquired by an additional sensor under the foot. Mean and standard deviation of minimum and maximum values were calculated for each interface area, locomotion task, and socket design. The mean pressure patterns during different locomotion tasks were reported, as well. Considering all subjects irrespective of socket design, the mean pressure range resulted 45.3 (posterior)-106.7 (posterior) kPa in horizontal walking; 48.3 (posterior)-113.8 (posterior) kPa in ascent walking; 50.8 (posterior)-105.7 (posterior) kPa in descent walking; 47.9 (posterior)-102.9 (lateral) kPa during upstairs; and 41.8 (posterior)-84.5 (anterior) kPa during downstairs. Qualitative differences in socket designs have been found. These data allow for a comprehensive analysis of pressures acting at the tissue-socket interface in people with transfemoral amputation, thus offering essential information for the design of novel solutions or to improve existing ones, in this field.

Closed-Chain Inverse Dynamics for the Biomechanical Analysis of Manual Material Handling Tasks through a Deep Learning Assisted Wearable Sensor Network.

Despite the automatization of many industrial and logistics processes, human workers are still often involved in the manual handling of loads. These activities lead to many work-related disorders that reduce the quality of life and the productivity of aged workers. A biomechanical analysis of such activities is the basis for a detailed estimation of the biomechanical overload, thus enabling focused prevention actions. Thanks to wearable sensor networks, it is now possible to analyze human biomechanics by an inverse dynamics approach in ecological conditions. The purposes of this study are the conceptualization, formulation, and implementation of a deep learning-assisted fully wearable sensor system for an online evaluation of the biomechanical effort that an operator exerts during a manual material handling task. In this paper, we show a novel, computationally efficient algorithm, implemented in ROS, to analyze the biomechanics of the human musculoskeletal systems by an inverse dynamics approach. We also propose a method for estimating the load and its distribution, relying on an egocentric camera and deep learning-based object recognition. This method is suitable for objects of known weight, as is often the case in logistics. Kinematic data, along with foot contact information, are provided by a fully wearable sensor network composed of inertial measurement units. The results show good accuracy and robustness of the system for object detection and grasp recognition, thus providing reliable load estimation for a high-impact field such as logistics. The outcome of the biomechanical analysis is consistent with the literature. However, improvements in gait segmentation are necessary to reduce discontinuities in the estimated lower limb articular wrenches.

Compensating the NLoS Occlusion Errors of UWB for Pedestrian Localization With MIMU

The Ultra-Wideband (UWB) technology has become a competitive choice for indoor pedestrian localization due to its advantages in accuracy, compact size, reliability, power efficiency, and easy setup. However, the None-Line-of-Sight (NLoS) occlusion caused by the human body parts and obstacles in its propagation path may introduce significant location errors that limit its performance and field of application. To handle the NLoS occlusion errors and promote the localization accuracy for complex indoor environments, a Micro-Electro-Mechanical System (MEMS) Inertial Measurement Unit (MIMU)-assisted UWB localization system with NLoS occlusion error compensation techniques is provided. Firstly, an NLoS occlusion detection method using the built-in channel information of the UWB transceivers and the variation pattern of anchor-node distance is proposed. Secondly, an Extended Kalman Filter (EKF)-based attitude computation and a Zero-Velocity Update (ZUPT)-based continuous gait segmentation are introduced for inertial navigation during occlusion intervals. Then, the UWB and IMU parameters are integrated with an Unscented Kalman Filter (UKF) framework to compensate for the occlusion error and improve the accuracy and robustness. Finally, a battery-powered miniature wearable device is developed and the proposed techniques are verified with experimental studies. The results demonstrate the feasibility and effectiveness of the presented techniques, which provides a good reference for related Internet of Things (IoT) applications.

Human gait-labeling uncertainty and a hybrid model for gait segmentation.

Evaluate manual labeling uncertainty and introduce a hybrid stride detection and gait-event estimation model for autonomous, long-term, and remote monitoring. Estimate inter-labeler inconsistencies by computing the limits-of-agreement. Develop a hybrid model based on dynamic time warping and convolutional neural network to identify valid strides and eliminate non-stride data in inertial (walking) data collected by a wearable device. Finally, detect gait events within a valid stride region. The limits of inter-labeler agreement for key gait events heel off, toe off, heel strike, and flat foot are 72, 16, 24, and 80 ms, respectively; The hybrid model's classification accuracy for stride and non-stride are 95.16 and 84.48%, respectively; The mean absolute error for detected heel off, toe off, heel strike, and flat foot are 24, 5, 9, and 13 ms, respectively, when compared to the average human labels. The results show the inherent labeling uncertainty and the limits of human gait labeling of motion capture data; The proposed hybrid-model's performance is comparable to that of human labelers, and it is a valid model to reliably detect strides and estimate the gait events in human gait data. This work establishes the foundation for fully automated human gait analysis systems with performances comparable to human-labelers.

A Method for Autonomous Multi-Motion Modes Recognition and Navigation Optimization for Indoor Pedestrian.

The indoor navigation method shows great application prospects that is based on a wearable foot-mounted inertial measurement unit and a zero-velocity update principle. Traditional navigation methods mainly support two-dimensional stable motion modes such as walking; special tasks such as rescue and disaster relief, medical search and rescue, in addition to normal walking, are usually accompanied by running, going upstairs, going downstairs and other motion modes, which will greatly affect the dynamic performance of the traditional zero-velocity update algorithm. Based on a wearable multi-node inertial sensor network, this paper presents a method of multi-motion modes recognition for indoor pedestrians based on gait segmentation and a long short-term memory artificial neural network, which improves the accuracy of multi-motion modes recognition. In view of the short effective interval of zero-velocity updates in motion modes with fast speeds such as running, different zero-velocity update detection algorithms and integrated navigation methods based on change of waist/foot headings are designed. The experimental results show that the overall recognition rate of the proposed method is 96.77%, and the navigation error is 1.26% of the total distance of the proposed method, which has good application prospects.

Adaptive Algorithm for Gait Segmentation Using a Single IMU in the Thigh Pocket

Gait segmentation may help monitoring the evolution of patients during rehabilitation treatment through the analysis of properly defined metrics. Many algorithms detect gait events using Inertial Measurement Units (IMUs). However, most of them require the IMUs to be attached to the body, or to assistive devices. Often individuals must go to a gait laboratory which owns specialized equipment so that gait analysis can be performed. IMUs are present in modern smartphones. An IMU<inline-formula> <tex-math notation="LaTeX">$/$ </tex-math></inline-formula>smartphone carried in the pocket during daily activities could allow the analysis of much more data in a more comfortable setting. We address the detection of four gait events: Heel-Strike (HS), Flat-Foot (FF), Heel-Off (HO) and Toe-Off (TO) using a single noisy IMU attached to a smartphone placed inside the thigh pocket. Gait was modelled as a four-state left-right Hidden Markov Model (HMM) whose observations follow multivariate Gaussian distributions. The decoding in post-processing was performed with a modified Viterbi decoder that accounts for a rule-based (RB) detection of TO. To validate the approach, experiments were performed in nine subjects with no gait abnormalities. Our algorithm obtained median F1-Scores <inline-formula> <tex-math notation="LaTeX">$\geq0.955$ </tex-math></inline-formula> for all events in intra-subject evaluation and median F1-Scores <inline-formula> <tex-math notation="LaTeX">$\geq0.757$ </tex-math></inline-formula> in inter-subject evaluations (with fast training and with a populational model). It demonstrated high generalization capability in our dataset and competitive performances when compared to other algorithms using high-quality IMUs attached to the body. This work is a step towards accurate and refined gait segmentation using a smartphone carried in the pocket.

Validity of Dual-Minima Algorithm for Heel-Strike and Toe-Off Prediction for the Amputee Population

Assessment of gait deficits relies on accurate gait segmentation based on the key gait events of heel strike (HS) and toe-off (TO). Kinematics-based estimation of gait events has shown promise in this regard, especially using the leg velocity signal and gyroscopic sensors. However, its validation for the amputee population is not established in the literature. The goal of this study is to assess the accuracy of lower-leg angular velocity signal in determining the TO and HS instants for the amputee population. An open data set containing marker data of 10 subjects with unilateral transfemoral amputation during treadmill walking was used. A rule-based dual-minima algorithm was developed to detect the landmarks in the shank velocity signal indicating TO and HS events. The predictions were compared against the force platform data for 2595 walking cycles from 239 walking trials. The results showed considerable accuracy for the HS with a median error of −1 ms. The TO prediction error was larger with the median ranging from 35–84 ms. The algorithm consistently predicted the TO earlier than the actual event. Significant differences were found between the prediction accuracy for the sound and prosthetic legs. The prediction accuracy was also affected by the subjects’ mobility level (K-level) but was largely unaffected by gait speed. In conclusion, the leg velocity profile during walking can predict the heel-strike and toe-off events for the transfemoral amputee population with varying degrees of accuracy depending upon the leg side and the amputee’s functional ability level.

Adaptive Pedestrian Stride Estimation for Localization: From Multi-Gait Perspective.

Accurate and reliable stride length estimation modules play a significant role in Pedestrian Dead Reckoning (PDR) systems, but the accuracy of stride length calculation suffers from individual differences. This paper presents a stride length prediction strategy for PDR systems that can be adapted across individuals and broad walking velocity fields. It consists of a multi-gait division algorithm, which can divide a full stride into push-off, swing, heel-strike, and stance based on multi-axis IMU data. Additionally, based on the acquired gait phases, the correlation between multiple features of distinct gait phases and the stride length is analyzed, and multi regression models are merged to output the stride length value. In experimental tests, the gait segmentation algorithm provided gait phases division with the F-score of 0.811, 0.748, 0.805, and 0.819 for stance, push-off, swing, heel-strike, respectively, and IoU of 0.482, 0.69, 0.509 for push-off, swing, heel-strike, respectively. The root means square error (RMSE) of our proposed stride length estimation was 151.933, and the relative error for total distance in varying walking speed tests was less than 2%. The experimental results validated that our proposed gait phase segmentation algorithm can accurately recognize gait phases for individuals with wide walking speed ranges. With no need for parameter modification, the stride length method based on the fusion of multiple predictions from different gait phases can provide better accuracy than the estimations based on the full stride.

Gait and Respiration-Based User Identification Using Wi-Fi Signal

The ever-growing security issues in various scenarios create an urgent demand for a reliable and convenient identification system. Traditional identification systems request users to provide passwords, fingerprints, or other easily stolen information. Existing works show that everyone’s gait and respiration have unique characteristics and are difficult to imitate. But these works only use gait or respiration information to achieve identification, which leads to low accuracy or long identification time. And they have no strong anti-interference ability, which leads to the limitation in practical application. Toward this end, we propose a new system which uses both gait and respiratory biometric characteristics to achieve user identification using Wi-Fi (GRi-Fi) in the presence of interferences. In our system, we design a segmentation algorithm to segment gait and respiration data. And we design a weighted subcarrier screening method to improve the anti-interference ability. In order to shorten the identification time, we propose a feature integration method based on the weighted average. Finally, we use a deep learning method to identify users accurately. Experimental results show that GRi-Fi can identify the users identity with an average accuracy of 98.3% in noninterference environments. Even in the presence of multiple interferences, the average identification accuracy also reaches 91.2%. In future applications, our system can be applied to many fields of Internet of Things, such as smart home systems and clocking in at companies.

The Diverse Gait Dataset: Gait Segmentation Using Inertial Sensors for Pedestrian Localization with Different Genders, Heights and Walking Speeds.

Stride length estimation is one of the most crucial aspects of Pedestrian Dead Reckoning (PDR). Due to the measurement noise of inertial sensors, individual variances of pedestrians, and the uncertainty in pedestrians walking, there is a substantial error in the assessment of stride length, which causes the accumulated deviation of Pedestrian Dead Reckoning (PDR). With the help of multi-gait analysis, which decomposes strides in time and space with greater detail and accuracy, a novel and revolutionary stride estimating model or scheme could improve the performance of PDR on different users. This paper presents a diverse stride gait dataset by using inertial sensors that collect foot movement data from people of different genders, heights, and walking speeds. The dataset contains 4690 walking strides data and 19,083 gait labels. Based on the dataset, we propose a threshold-independent stride segmentation algorithm called SDATW and achieve an F-measure of 0.835. We also provide the detailed results of recognizing four gaits under different walking speeds, demonstrating the utility of our dataset for helping train stride segmentation algorithms and gait detection algorithms.

A Novel Algorithm for Automatic Estimation of Gait Parameters Using Lower-Body Kinematics

Accurate estimation of gait parameters from the limb kinematics data remains a challenge for researchers. Several deterministic algorithms have been proposed in the literature for this purpose but their applicability is limited by their accuracy, especially for gait-impaired subjects. Gait parameter estimation relies on the accurate detection of key gait events of initial contact (IC) and terminal contact (TC) for gait segmentation. This study aimed to evaluate literature algorithms for IC and TC prediction for the amputee population and to introduce a novel algorithm to improve gait parameter estimation. A marker data set of 12 amputee subjects during treadmill walking was used. Seven algorithms, including one novel algorithm, detecting the landmarks (maxima, minima, and zero-crossings) in the foot, shank, and thigh velocity data were implemented and their IC and TC estimations were compared against the force platform data. IC was best detected by foot and shank velocity methods with median bias in the range of -3 to -27ms. However, the literature algorithms produced inferior accuracy for the TC event with consistent early prediction. The newly-proposed thigh velocity-based algorithm produced the best result for the TC prediction -14 to 19ms median bias. By combining the IC prediction from foot velocity data and the TC prediction from the thigh velocity, the algorithm produced the best results for temporal gait parameters with a median bias of within 5% of gait cycle time. Statistical analysis revealed the robustness of the algorithm across most walking speeds.

Ambulatory Human Gait Phase Detection Using Wearable Inertial Sensors and Hidden Markov Model.

Gait analysis, as a common inspection method for human gait, can provide a series of kinematics, dynamics and other parameters through instrumental measurement. In recent years, gait analysis has been gradually applied to the diagnosis of diseases, the evaluation of orthopedic surgery and rehabilitation progress, especially, gait phase abnormality can be used as a clinical diagnostic indicator of Alzheimer Disease and Parkinson Disease, which usually show varying degrees of gait phase abnormality. This research proposed an inertial sensor based gait analysis method. Smoothed and filtered angular velocity signal was chosen as the input data of the 15-dimensional temporal characteristic feature. Hidden Markov Model and parameter adaptive model are used to segment gait phases. Experimental results show that the proposed model based on HMM and parameter adaptation achieves good recognition rate in gait phases segmentation compared to other classification models, and the recognition results of gait phase are consistent with ground truth. The proposed wearable device used for data collection can be embedded on the shoe, which can not only collect patients’ gait data stably and reliably, ensuring the integrity and objectivity of gait data, but also collect data in daily scene and ambulatory outdoor environment.

Lower Limb Model Based Inertial Indoor Pedestrian Navigation System for Walking and Running

Foot-mounted inertial navigation systems are suitable in GNSS denied environment for special tasks such as fire rescue, soldier, etc. These tasks are often accompanied by vigorous movements such as running and jumping, which is a challenge for the Zero Velocity Update (ZUPT) based inertial navigation system. As the speed increases, the inertial sensor drift increases, while the zero-velocity interval shortens, making the gait detection more difficult and resulting in a limited error correction. And as the motion intensifies, the stability of the system decreases. A multi-node inertial pedestrian navigation system based on the lower limb model constraint is proposed in this paper. Our first contribution in this paper is studying a simplified adaptive zero-bias estimation algorithm. The changes in velocity and angular velocity between two steps are used to estimate and compensate for the drift of accelerometers and gyroscopes. Our second contribution is proposing a novel gait segmentation method by lower limb motion statistics feature. A hybrid detection model of dual IMU is conducted to improve the accuracy of gait recognition. After that, the maximum distance inequality constraint is studied to correct the position error when the still phase is not available in vigorous movement. The final contribution is studying the chi-square fault detection method to improve the stability of KF. Multiple experiments prove that the proposed multi-node inertial pedestrian navigation system is effective in dealing with various challenging in walking, running, and hybrid motion pedestrian inertial navigation problems which could achieve less than 0.3% average distance error.

Gait Segmentation Method Using a Plantar Pressure Measurement System with Custom-Made Capacitive Sensors.

Gait analysis has been widely studied by researchers due to the impact in clinical fields. It provides relevant information on the condition of a patient’s pathologies. In the last decades, different gait measurement methods have been developed in order to identify parameters that can contribute to gait cycles. Analyzing those parameters, it is possible to segment and identify different phases of gait cycles, making these studies easier and more accurate. This paper proposes a simple gait segmentation method based on plantar pressure measurement. Current methods used by researchers and clinicians are based on multiple sensing devices (e.g., multiple cameras, multiple inertial measurement units (IMUs)). Our proposal uses plantar pressure information from only two sensorized insoles that were designed and implemented with eight custom-made flexible capacitive sensors. An algorithm was implemented to calculate gait parameters and segment gait cycle phases and subphases. Functional tests were performed in six healthy volunteers in a 10 m walking test. The designed in-shoe insole presented an average power consumption of 44 mA under operation. The system segmented the gait phases and sub-phases in all subjects. The calculated percentile distribution between stance phase time and swing phase time was almost 60%/40%, which is aligned with literature reports on healthy subjects. Our results show that the system achieves a successful segmentation of gait phases and subphases, is capable of reporting COP velocity, double support time, cadence, stance phase time percentage, swing phase time percentage, and double support time percentage. The proposed system allows for the simplification of the assessment method in the recovery process for both patients and clinicians.

Estimation of Stride Time Variability in Unobtrusive Long-Term Monitoring Using Inertial Measurement Sensors.

Stride time variability is an important indicator for the assessment of gait stability. An accurate extraction of the stride intervals is essential for determining stride time variability. Peak detection is a commonly used method for gait segmentation and stride time estimation. Standard peak detection algorithms often fail due to additional movement components and measurement noise. A novel algorithm for robust peak detection in inertial sensor signals was proposed in a previous contribution. In this work, we present a novel approach for estimation of stride time variability based on the formerly proposed peak detection algorithm applied to an unobtrusive sensor setup for motion monitoring. The unobtrusive sensor setup includes a wrist sensor, a pocket or belt sensor, and a necklace sensor, all equipped with both accelerometer and gyroscope. The goal of this work is to implement a generalized approach for accurate and robust stride interval determining algorithm for different sensor locations. Therefore, treadmill and level ground walking experiments were conducted with ten healthy subjects at increasing walking speeds and an age-simulating suit. With the proposed algorithm, we achieved a RMSE of 0.07s for the stride interval estimation during treadmill walking experiments. The results give promising indications that detection of variation of stride time variability is possible using the proposed unobtrusive sensor setup.

A Survey of Teleceptive Sensing for Wearable Assistive Robotic Devices.

Teleception is defined as sensing that occurs remotely, with no physical contact with the object being sensed. To emulate innate control systems of the human body, a control system for a semi- or fully autonomous assistive device not only requires feedforward models of desired movement, but also the environmental or contextual awareness that could be provided by teleception. Several recent publications present teleception modalities integrated into control systems and provide preliminary results, for example, for performing hand grasp prediction or endpoint control of an arm assistive device; and gait segmentation, forward prediction of desired locomotion mode, and activity-specific control of a prosthetic leg or exoskeleton. Collectively, several different approaches to incorporating teleception have been used, including sensor fusion, geometric segmentation, and machine learning. In this paper, we summarize the recent and ongoing published work in this promising new area of research.

GaitNet: An end-to-end network for gait based human identification

Gait recognition is one of the most important techniques for human identification at a distance. Most current gait recognition frameworks consist of several separate steps: silhouette segmentation, feature extraction, feature learning, and similarity measurement. These modules are mutually independent with each part fixed, resulting in a suboptimal performance in challenging conditions. In this paper, we integrate those steps into one framework, i.e., an end-to-end network for gait recognition, named GaitNet. It is composed of two convolutional neural networks: one corresponds to gait segmentation, and the other corresponds to classification. The two networks are modeled in one joint learning procedure which can be trained jointly. This strategy greatly simplifies the traditional step-by-step manner and is thus much more efficient for practical applications. Moreover, joint learning can automatically adjust each part to fit the global optimal objective, leading to obvious performance improvement over separate learning. We evaluate our method on three large scale gait datasets, including CASIA-B, SZU RGB-D Gait and a newly built database with complex dynamic outdoor backgrounds. Extensive experimental results show that the proposed method is effective and achieves the state-of-the-art results. The code and data will be released upon request.