An Easily Customized Gesture Recognizer for Assisted Living Using Commodity Mobile Devices

Mezari, Antigoni

doi:10.1155/2018/3180652

Cited by 11 publications

(7 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Related to our work, uWave [51] stored templates of accelerometer signals for new gestures, and used DTW to compare against incoming data streams. Bigdelou et al [7] applied Laplacian Eigenmaps and kernel regression on arm-worn IMU signals, while Mezari et al [58] leveraged fast Fourier transforms (FFT), symbolic aggregate approximation, and simple distance metrics to recognize new gestures.…”

Section: Gesture Customizationmentioning

confidence: 99%

Enabling hand gesture customization on wrist-worn devices

Xu,

Gong,

Brum

et al. 2022

Preprint

View full text Add to dashboard Cite

We present a framework for gesture customization requiring minimal examples from users, all without degrading the performance of existing gesture sets. To achieve this, we first deployed a large-scale study (N=500+) to collect data and train an accelerometer-gyroscope recognition model with a cross-user accuracy of 95.7% and a falsepositive rate of 0.6 per hour when tested on everyday non-gesture data. Next, we design a few-shot learning framework which derives a lightweight model from our pre-trained model, enabling knowledge transfer without performance degradation. We validate our approach through a user study (N=20) examining on-device customization from 12 new gestures, resulting in an average accuracy of 55.3%, 83.1%, and 87.2% on using one, three, or five shots when adding a new gesture, while maintaining the same recognition accuracy and false-positive rate from the pre-existing gesture set. We further evaluate the usability of our real-time implementation with a user experience study (N=20). Our results highlight the effectiveness, learnability, and usability of our customization framework.

show abstract

Section: Gesture Customizationmentioning

confidence: 99%

Enabling hand gesture customization on wrist-worn devices

Xu,

Gong,

Brum

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…The three membership functions µ2, µ3 and µ4 are superimposed and merged, and the recognition accuracy of the proposed algorithm is retested using the fused membership function. The algorithm is compared with the four motion recognition algorithms supporting small amplitude gesture recognition mentioned in literature [36] and literature [37], and the recognition result and ROC curve are shown in Figure 18 and Figure 19. It can be seen from the results that the algorithm combines motion information and RGB three-color features, and shows a high recognition accuracy for small-scale motion, which is significantly better than the first three algorithms.…”

Section: Performance Comparison Of Two Fusion Strategiesmentioning

confidence: 99%

Intelligent Recognition of Medical Motion Image Combining Convolutional Neural Network With Internet of Things

Zhou

Gao

2019

IEEE Access

View full text Add to dashboard Cite

For the small-scale motion in medical motion images, the traditional medical motion image intelligent recognition algorithm has low recognition accuracy, and requires a large amount of calculation statistics. There is no self-learning function, which seriously affects the accuracy and speed of medical motion image recognition. Therefore, in order to improve the accuracy of human body small-scale motion recognition in medical motion images and the computational efficiency of large-scale data sets, an intelligent recognition algorithm based on convolutional neural network for medical motion images is proposed. The algorithm first learns the dense trajectory features and depth features, and then further fuses the dense trajectory features with the deep learning features. Finally, the extreme learning machine is applied to the convolutional neural network, and the fused features are further trained as input information of the convolutional neural network, and the features from the bottom layer to the upper layer can be extracted step by step from the raw data of the pixel level. Simulation experiments show that the algorithm can effectively improve the recognition accuracy of small-scale motion in medical moving images and improve the speed of motion.INDEX TERMS Internet of things, convolutional neural network, medical motion image, intelligent recognition.

show abstract

“…Human motion analysis is a methodology used in many fields [2,10,48], from athletic performance analysis and biomechanical representation of the musculoskeletal system, to consumer and entertainment applications such as exergames [36,37]. Core elements of the respective techniques are typically advanced sensors such as optoelectronic systems, ultrasonic localization systems, electromagnetic measurement systems [47], gyroscopes [49], depth cameras, accelerometers [26], or other special equipment and platforms that the users are required to operate. The common characteristic of the above technologies and hardware is that they are very expensive in purchase and operation, and the overall system is complex, with limited extensibility capabilities.…”

Section: Introductionmentioning

confidence: 99%

On unifying deep learning and edge computing for human motion analysis in exergames development

Pardos¹,

Menychtas²

2021

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

This work describes a novel methodology for creating exergames on an edge-native platform with the integration of multiple deep neural networks. A prototype of the platform, which includes capabilities for innovative gameplay and advanced user interactivity, has been implemented and deployed in a real-world scenario. At core of the proposed methodology is the ad hoc training of classifiers for posture classification which can be dynamically adapted to the specific requirements of the usage scenario, operational and environmental conditions allowing for real-time identification of events and advanced game control. The proposed solution is ideal for individual consumers in a home environment since is supports by-design edge platforms minimizing the cost of the system and enabling in parallel the communication with state-of-the-art hardware (i.e., GPUs, TPUs, computer boards) for real-time operation. The proposed system allows the collection and analysis of game data, which can be exploited by specialized personnel in rehabilitation centers or for other purposes in the areas of healthcare and assisted living.

show abstract

An Easily Customized Gesture Recognizer for Assisted Living Using Commodity Mobile Devices

Cited by 11 publications

References 35 publications

Enabling hand gesture customization on wrist-worn devices

Enabling hand gesture customization on wrist-worn devices

Intelligent Recognition of Medical Motion Image Combining Convolutional Neural Network With Internet of Things

On unifying deep learning and edge computing for human motion analysis in exergames development

Contact Info

Product

Resources

About