Feature covariance matrix-based dynamic hand gesture recognition

Fang, Linpu; Wu, Guile; Kang, Wenxiong; Wu, Qiuxia; Wang, Zhiyong; Feng, David Dagan

doi:10.1007/s00521-018-3719-3

Cited by 7 publications

(3 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Modifications and extensions to the dataset used allow the created model to be generalized, making it less sensitive to changes in the environment. Approaches with relatively limited generalization capability include methods such as support vector machines [8], [9] and hidden Markov models [10]. With the development and successful application of deep learning in the field of image recognition, recent work in hand detection has mainly focused on convolutional network models, which have enabled to significantly increase the accuracy of image segmentation methods [11].…”

Section: Related Work a Hand Localizationmentioning

confidence: 99%

Generating Synthetic Depth Image Dataset for Industrial Applications of Hand Localization

et al. 2022

View full text Add to dashboard Cite

In this paper, we focus on the problem of applying domain randomization to produce synthetic datasets for training depth image segmentation models for the task of hand localization. We provide new synthetic datasets for industrial environments suitable for various hand tracking applications, as well as ready-to-use pre-trained models. The presented datasets are analyzed to evaluate the characteristics of these datasets that affect the generalizability of the trained models, and recommendations are given for adapting the simulation environment to achieve satisfactory results when creating datasets for specialized applications. Our approach is not limited by the shortcomings of standard analytical methods, such as color, specific gestures, or hand orientation. The models in this paper were trained solely on a synthetic dataset and were never trained on real camera images; nevertheless, we demonstrate that our most diverse datasets allow the models to achieve up to 90% accuracy. The proposed hand localization system is designed for industrial applications where the operator shares the workspace with the robot.

show abstract

Section: Related Work a Hand Localizationmentioning

confidence: 99%

Generating Synthetic Depth Image Dataset for Industrial Applications of Hand Localization

et al. 2022

View full text Add to dashboard Cite

show abstract

“…This method can recognise similar gestures accurately, but it needs depth data and does not apply to mobile terminals. In [11], the authors proposed a novel method using the feature covariance matrix which can be applied to the two‐dimensional (2D), depth‐based and skeleton‐based 3D datasets. However, the shortcoming of this method is it cannot recognise multiple hands.…”

Section: Related Workmentioning

confidence: 99%

“…Apart from that, there are also some deep learning‐based algorithms used for dynamic gesture recognition. They use the convolutional neural network (CNN) [11] to extract image features and use long short‐term memory (LSTM) [12, 13] to classify the dynamic gesture.…”

Section: Introductionmentioning

confidence: 99%