Fig 1 - uploaded by Ariful Islam Anik
Content may be subject to copyright.
Source publication
The scope for doing physical exercises in daily life is declining day by day. But, the importance of human physical exercise for a healthy life, remains the same. It is necessary to generate a solution to simulate the outdoor experience of physical exercises and sports inside our home. In this paper, we propose an idea of recognizing the activities...
Similar publications
Team handball is a fast and complex game with a very traditional background and so far, almost no collection of digital information. Only a few attempts have been made to come up with models to explain the mechanisms of the game based on measured indicators. CoCoAnDa is a project located at the Baden-Wuerttemberg Cooperative State University that a...
Citations
... Anik et al. [20] placed a sensor including accelerometer and gyroscope in the racket and derived a set of 8 features from each axis (for a total of 48 features), to then train an SVM classifier for 3 types of stroke (smash, serve, backhand). However, the size of the dataset was limited to 180 instances, and the sensor was placed in the cords of the racket, which might hinder its handling when playing a real game. ...
... Human activity recognition can assist to automatically save energy in smart homes, such as heating, ventilation, air conditioning and lighting by understanding user's intentions [1], [2]. Several sensor modalities for human activity recognition are investigated in literature such as camera based, radar [3]- [7], Inertial Measurement Unit (IMU) [8]- [10], infrared, thermal imaging sensors, etc. The machine learning models used for HAR are optimized using categorical distribution based loss functions such as softmax [3], [11], [12]. ...
Human activity recognition (HAR) using IMU sensors, namely accelerometer and gyroscope, has several applications in smart homes, healthcare and human-machine interface systems. In practice, the IMU-based HAR system is expected to encounter variations in measurement due to sensor degradation, alien environment or sensor noise and will be subjected to unknown activities. In view of practical deployment of the solution, analysis of statistical confidence over the activity class score are important metrics. In this paper, we therefore propose XAI-BayesHAR, an integrated Bayesian framework, that improves the overall activity classification accuracy of IMU-based HAR solutions by recursively tracking the feature embedding vector and its associated uncertainty via Kalman filter. Additionally, XAI-BayesHAR acts as an out of data distribution (OOD) detector using the predictive uncertainty which help to evaluate and detect alien input data distribution. Furthermore, Shapley value-based performance of the proposed framework is also evaluated to understand the importance of the feature embedding vector and accordingly used for model compression
... Contrastingly, we mount sensors on four different limb positions (wrist, palm, left leg and right leg) and consider an increased number of strokes compared to the above study. In [12], the authors develop a classifier model that can detect badminton strokes. The data is collected using a Magnetic Pickup Unit (MPU 6050 sensor), and an Arduino module is employed as a hub. ...
... • Sports Analytics in AI Domain: Recently, researchers have proposed novel approaches and techniques in sensorbased and camera-based applications in sports analytics covering a vast spectrum of topics ranging from activity classification [12], segmentation techniques [17,18], virtual reality console [19] and sports training decision support system [20]. Moreover, nowadays, researchers are investigating to understand the dynamics behind the game, which will help them postulate novel approaches to assist the players in improving their performances. ...
Wearable devices have gained immense popularity among various pervasive computing and Internet-of-Things (IoT) applications in the past decade. Sports analytics researchers recently focused on improving a player’s performance to help devise a winning strategy based on the player’s gameplay. Especially in a racquet-based badminton sport, it is assumed that handling the racquet during the gameplay is one of the primary reasons to influence the players’ performance. On the contrary, we posit that the players’ stance, body movements, and posture are equally significant in evaluating a player’s performance during the game. A shot characterized by a recommended posture, stance, and body movements allows a player to play a stroke efficiently, thus aiding the player in guiding the shuttle to strategic spots and making it difficult for the opponent to return the shot and score a point. Relying on this hypothesis, we propose DeCoach, a data-driven framework that leverages the stance and posture of the players and ranks them based on their performances. In this effort, we first employ a deep learning-based algorithm to classify the strokes and stances of the players. Secondly, we propose a distance-based methodology to compare the obtained stance of a player with that of a professional player. Finally, we devise a deep learning-based regressor to predict the player’s performance which commences with ranking based on their performance. We evaluate DeCoach using our in-house dataset, Badminton Activity Recognition (BAR) Dataset that is collected using inertial measurement unit (IMU) sensors by placing them on the upper and lower limbs of the players. The BAR dataset is collected from 11 players in the controlled and uncontrolled environment settings for 12 frequently played shots in the game. Empirical results indicate that DeCoach achieves 89.09% accuracy for strokes detection and R2 score of 88.84% in estimating the players’ performance.
... HAR data are fed into algorithms, thereby allowing goals of monitoring, analysis, and assisting humans to be achieved [1]. For example, in fields such as sports training, medicine, and motion sensing [2][3][4][5][6][7][8][9][10][11], various sensors have been used to collect data on human activities, for example, from human computer interaction to surveillance, security, and health monitoring systems. Despite ongoing efforts in the field, the research addresses that activity recognition is still a difficult task in an unrestricted environment and faces many challenges [12]. ...
This paper presents a wearable device, fitted on the waist of a participant that recognizes six activities of daily living (walking, walking upstairs, walking downstairs, sitting, standing, and laying) through a deep-learning algorithm, human activity recognition (HAR). The wearable device comprises a single-board computer (SBC) and six-axis sensors. The deep-learning algorithm employs three parallel convolutional neural networks for local feature extraction and for subsequent concatenation to establish feature fusion models of varying kernel size. By using kernels of different sizes, relevant local features of varying lengths were identified, thereby increasing the accuracy of human activity recognition. Regarding experimental data, the database of University of California, Irvine (UCI) and self-recorded data were used separately. The self-recorded data were obtained by having 21 participants wear the device on their waist and perform six common activities in the laboratory. These data were used to verify the proposed deep-learning algorithm on the performance of the wearable device. The accuracy of these six activities in the UCI dataset and in the self-recorded data were 97.49% and 96.27%, respectively. The accuracies in tenfold cross-validation were 99.56% and 97.46%, respectively. The experimental results have successfully verified the proposed convolutional neural network (CNN) architecture, which can be used in rehabilitation assessment for people unable to exercise vigorously.
... Proactively sensing the user's activity can enable smart automatic control of devices to save energy demand in smart homes, such as heating, ventilation and air conditioning and lighting [1], [2]. Among other modalities, such as camerabased solution, radar [3]- [8], accelerometer-gyroscope IMU sensors [9]- [11] are predominantly used for humans activity classification. Such sensors are able to register tiny movements, which typically remains elusive in other sensing modalities. ...
... A patent from Fisher [2] describes pole forces and angles relative to the ground using force sensors and the inertial measurement units (IMUs); however, this method provides no physical way to present data. In other sports, sensors integrated into sports equipment have captured movements [3] with high accuracy and recognized specific movements [4], [5]; integrating sensors for smart sports equipment [6] makes research and understanding of technique possible and creates the foundation for product development. ...
A modular sensor application for measuring athlete performance in skiing sports was developed. Using inertial measurement units (IMUs) and load cells in a modular system, a force orientation measurement system, FOMS, was developed. A functioning prototype capable of measuring ski sports dynamics was created. Data processing using the system, a validation of the prototype in terms of angle measurement IMU accuracy, example data from in-field athlete testing, and visualization by animations are described. The system developed contains four subsystems: a controller, two pole measuring modules, and a terrain-measuring module. The system structure also allows for additional modules, making the system applicable to different sports. The IMUs use orientation-sensing components to measure pole orientations, which are used to calculate decomposed forces relative to the terrain. Data from different modules are synchronized using wireless communication and saved on SD cards with time stamps. A validation experiment was conducted in which the angles from the modules were compared with the Oqus motion capture system from Qualisys. Examples for athlete testing in both cross country and alpine skiing were calculated from the matrix provided by the different modules and are presented in graphs to evaluate the athlete. In addition, the relative pole/terrain coordinates are visualized in 2D and 3D animations for analyzing the movement pattern in connection with the applied forces, opening up a whole new level of sports analysis.
... In the domain of racket sports, which typically involves more complex techniques and dynamic movement, Sharma et al. [21] present a technique based on a combination of Inertial Measurement Unit and audio sensor data embedded in a smartwatch to detect various table tennis strokes. Anik et al. [22] used a motion-tracking device that contains both accelerometer and gyroscope attached to the badminton racket to recognise the different types of badminton strokes such as serve, smash, backhand, forehand, return, etc. The results from these three papers show that the combination of sensors contributes to a better recognition rate and improvement than using these sensors individually. ...
... In another supervised learning method example, Anik et al. [22] applied two machine learning model approaches that are the K-Nearest Neighbors and Support Vector Machines (SVM) classifiers using the motion data collected from a large set of users to recognise the different types of badminton strokes. Similarly, Blank et al. [26] presented an approach for table tennis stroke detection and stroke type classification using inertial sensors attached to table tennis rackets and collected data of eight different basic stroke types from ten amateur and professional players. ...
Beginner table-tennis players require constant real-time feedback while learning the fundamental techniques. However, due to various constraints such as the mentor's inability to be around all the time, expensive sensors and equipment for sports training, beginners are unable to get the immediate real-time feedback they need during training. Sensors have been widely used to train beginners and novices for various skills development, including psychomotor skills. Sensors enable the collection of multimodal data which can be utilised with machine learning to classify training mistakes, give feedback, and further improve the learning outcomes. In this paper, we introduce the Table Tennis Tutor (T3), a multi-sensor system consisting of a smartphone device with its built-in sensors for collecting motion data and a Microsoft Kinect for tracking body position. We focused on the forehand stroke mistake detection. We collected a dataset recording an experienced table tennis player performing 260 short forehand strokes (correct) and mimicking 250 long forehand strokes (mistake). We analysed and annotated the multimodal data for training a recurrent neural network that classifies correct and incorrect strokes. To investigate the accuracy level of the aforementioned sensors, three combinations were validated in this study: smartphone sensors only, the Kinect only, and both devices combined. The results of the study show that smartphone sensors alone perform sub-par than the Kinect, but similar with better precision together with the Kinect. To further strengthen T3's potential for training, an expert interview session was held virtually with a table tennis coach to investigate the coach's perception of having a real-time feedback system to assist beginners during training sessions. The outcome of the interview shows positive expectations and provided more inputs that can be beneficial for the future implementations of the T3.
... Thus, the created feature vector which consists of 18 elements is passed into the algorithmic model to detect activity. Support Vector Machines (SVM) has been used previously in different types of activity recognition [14] [6]. SVM is applied to the dataset.80% is used as the training set and 20% is used as the test set. ...
... Untuk mengklasifikasikan data, dibutuhkan fitur-fitur tertentu sebagai dasar penentuan kelas. Pada kasus pendeteksi aktivitas manusia digunakan fitur-fitur pada Tabel 2.[1] Tabel 2. Fitur-Fitur Dari Data Gerakan Fitur Deskripsi Rata-rata ...
Internet of Things (IoT) dapat diaplikasikan untuk banyak bidang, salah satunya pada latihan olahraga bulu tangkis. Pada olahraga bulu tangkis, terutama bagi pemain pemula mengalami kesulitan untuk mengetahui apakah gerakan yang dilakukan sudah benar atau belum. Pada penelitian ini, dibangun sebuah embedded system yang dipasang pada raket yang berfungsi mengambil data gerakan pukulan. Data pukulan ini dikirim ke sebuah perangkat lunak yang dapat mendeteksi jenis gerakan raket bulu tangkis. Embedded system terdiri dari Arduino dan sensor accelerometer dan gyroscope. Data pukulan disimpan ke dalam basis data melalui web service. Perangkat lunak dibangun dengan memanfaatkan prinsip pembelajaran mesin terarah yaitu klasifikasi. Algoritma klasifikasi yang digunakan adalah algoritma k-Nearest Neighbor dan membandingkan hasilnya dengan algoritma lain yaitu Support Vector Machine. Pengujian dilakukan dengan mengumpulkan data latih yang digunakan oleh algoritma klasifikasi untuk memprediksi gerakan. Kinerja dari kedua algoritma klasifikasi diukur dan dibandingkan. Dari hasil pengujian, maka disimpulkan bahwa algoritma Support Vector Machine menghasilkan kinerja yang lebih baik dari k-Nearest Neighbor dalam mendeteksi gerakan raket. Selain itu kinerja algoritma Support Vector Machine yang lebih baik tersebut dihasilkan dengan data latih yang lebih sedikit dibandingkan k-Nearest Neighbor.
... Today, the ubiquity of sensors (e.g., accelerometers, gyroscopes, and magnetometers) and their availability in mobile platforms make it easy to measure or analyze different aspects of physical activities e.g., motion, location and direction. The data collected by sensors are widely used to develop solutions in several domains such as healthcare [2], security [3], robotics [4], transportation [5], sports [6], smart home [7] and smart city [8]. Furthermore, HAR is one of the most assisting technology tools to support the elderly's daily life [9]. ...
... The learned models are then (5) validated by the test data frames. In the next step, performance measures are evaluated on the classification results (6), and finally the hybrid models are compared based on the evaluation results (7). The above steps are explained in the following sections. ...
Recent advances in artificial intelligence and machine learning (ML) have led to effective 1 methods and tools for analyzing the human behavior. Human Activity Recognition (HAR) is one 2 of the fields that has seen an explosive research interest among the ML community due to its wide 3 range of applications. HAR is one of the most assisting technology tools to support the elderly's 4 daily life and to help people suffering from cognitive disorders, Parkinson's disease, dementia, etc. 5 It is also very useful in areas such as transportation, robotics and sports. Deep learning (DL) is a 6 branch of ML based upon complex Artificial Neural Networks (ANNs) that has demonstrated a high 7 level of accuracy and performance in HAR. Convolutional Neural Networks (CNNs) and Recurrent 8 Neural Networks (RNNs) are two types of DL models widely used in the recent years to address 9 the HAR problem. The purpose of this paper is to investigate the effectiveness of their integration in 10 recognizing daily activities, e.g., walking. We analyze four hybrid models that integrate CNNs with 11 four powerful RNNs, i.e., LSTMs, BiLSTMs, GRUs and BiGRUs. The outcomes of our experiments 12 on the PAMAP2 dataset indicate that our proposed hybrid models achieve an outstanding level 13 of performance with respect to several indicative measures, e.g., F-score, accuracy, sensitivity and 14 specificity. 15
