Model transfer from 2D to 3D study for boxing pose estimation

Abstract

IntroductionBoxing as a sport is growing on Chinese campuses, resulting in a coaching shortage. The human pose estimation technology can be employed to estimate boxing poses and teach interns to relieve the shortage. Currently, 3D cameras can provide more depth information than 2D cameras. It can potentially improve the estimation. However, the input channels are inconsistent between 2D and 3D images, and there is a lack of detailed analysis about the key point location, which indicates the network design for improving the human pose estimation technology.MethodTherefore, a model transfer with channel patching was implemented to solve the problems of channel inconsistency. The differences between the key points were analyzed. Three popular and highly structured 2D models of OpenPose (OP), stacked Hourglass (HG), and High Resolution (HR) networks were employed. Ways of reusing RGB channels were investigated to fill up the depth channel. Then, their performances were investigated to find out the limitations of each network structure.Results and discussionThe results show that model transfer learning by the mean way of RGB channels patching the lacking channel can improve the average accuracies of pose key points from 1 to 20% than without transfer. 3D accuracies are 0.3 to 0.5% higher than 2D baselines. The stacked structure of the network shows better on hip and knee points than the parallel structure, although the parallel design shows much better on the residue points. As a result, the model transfer can practically fulfill boxing pose estimation from 2D to 3D.

3D human pose data augmentation using Generative Adversarial Networks for robotic-assisted movement quality assessment

“…GANs, with their unique architecture comprising a generator and a discriminator, are inherently suited for generating new, realistic data instances that mimic the distribution of real data. This capability is critical in our context (Lin et al, 2023). To substantiate our choice, we have conducted extensive comparative analysis, demonstrating that GANs outperform other models in generating realistic, diverse, and application-specific synthetic data for 3D human pose augmentation.…”

Boxing behavior recognition based on artificial intelligence convolutional neural network with sports psychology assistant

“…In order to evaluate the performance of the model in this study, the model algorithm in this study is compared with the model algorithm proposed by 3D-CNN 45 , 3D-ResNet, ResNet 46 and Zhang et al 32 , and evaluated from loss value, accuracy and F1 value respectively, as shown in Figs. 6 , 7 and 8 .…”