![]() It has a dynamic sample conversion and a heavy 2D pose analysis section. ![]() But CVF3D model has some problems and shortcomings. The performance of the CVF3D model is also better than Tri-CPM and AutoEnc. Its actual performance is better (Table 4). ![]() The multi-view fusion strategy in this model is a novel and long-acting optimization framework. The new cross-view fusion 3D human pose estimation model (CVF3D) generates human movements in three-dimensional space by fusing the multi-view 2D poses heatmap more accurately. Recently, 3D human pose estimation has become a very important practical task. In our work, we used these networks as experimental comparisons to reflect the superiority of the network designed in this paper in terms of network size and estimation performance. And these networks have room for improvement in pose estimation performance. The Resnet network has a relatively large number of parameters, while the Mobilenet and Efficientnetv2 networks are not satisfactory in terms of fast start-up. But these networks also have some disadvantages. Similarly, the structure of the Efficientnetv2 network is lighter. While, the Mobilenet series network uses the inverted residual to extract more refined features by expanding the dimension of the tensor. In the estimation of human pose, this Resnet series network is superior in training speed and effectiveness due to its residual network. The Resnet series network has already obtained mature applications in many fields. Grant Recipient:Ming-hui Sun.Ĭompeting interests: The authors have declared that no competing interests exist. 61272209, 61872164), in part by the Program of Science and Technology Development Plan of Jilin Province of China under Grant 20190302032GX, and in part by the Fundamental Research Funds for the Central Universities (Jilin University). This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.ĭata Availability: All relevant data are within the paper and its Supporting information files.įunding: This work was supported by the National Natural Science Foundation of China (Grant Nos. Received: JAccepted: FebruPublished: February 23, 2022Ĭopyright: © 2022 Wang et al. Compared with other pose estimation models, its performance has also reached a higher level of application.Ĭitation: Wang H, Sun M-h, Zhang H, Dong L-y (2022) LHPE-nets: A lightweight 2D and 3D human pose estimation model with well-structural deep networks and multi-view pose sample simplification method. The experiment analyzes the network size, fast start-up and the performance in 2D and 3D pose estimation of the model in this paper in detail. Its performance in the overall human pose estimation exceeds other networks by more than 7 mm. And they also show the accuracy improvement of this work in estimating different joints, the estimated performance of approximately 60% of the joints is improved. The experimental results show the superiority of this work in fast start-up and network lightweight, it is about 1-5 epochs faster than the Resnet-34 during training. In the experiment, we used several recent models and two public estimation indicators. It implemented low-cost sample storage, and it was also convenient for models to read these samples. And we also designed a static pose sample simplification method for 3D pose data. LHPE-nets uses a network structure with evenly distributed channels, inverted residuals, external residual blocks and a framework for processing small-resolution samples to achieve training saturation faster. We call our network in this article LHPE-nets, which mainly includes Low-Span network and RDNS network. Then, based on the performance and drawbacks of these networks, we built multiple deep learning networks with better performance. These deep networks include Mobilenetv2, Mobilenetv3, Efficientnetv2 and Resnet. Therefore, in this article, we tested some new deep learning networks for pose estimation tasks. It uses some deep learning networks to generate heatmaps for each view. The multi-view 2D pose estimation part of this model is very important, but its training cost is also very high. The cross-view 3D human pose estimation model has made significant progress, it better completed the task of human joint positioning and skeleton modeling in 3D through multi-view fusion method.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |