- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2018 International Conference on 3D Vision (3DV) - Verona (2018.9.5-2018.9.8)] 2018 International Conference on 3D Vision (3DV) - DeepHPS: End-to-end Estimation of 3D Hand Pose and Shape by Learning from Synthetic Depth
摘要： Articulated hand pose and shape estimation is an important problem for vision-based applications such as augmented reality and animation. In contrast to the existing methods which optimize only for joint positions, we propose a fully supervised deep network which learns to jointly estimate a full 3D hand mesh representation and pose from a single depth image. To this end, a CNN architecture is employed to estimate parametric representations i.e. hand pose, bone scales and complex shape parameters. Then, a novel hand pose and shape layer, embedded inside our deep framework, produces 3D joint positions and hand mesh. Lack of sufficient training data with varying hand shapes limits the generalized performance of learning based methods. Also, manually annotating real data is suboptimal. Therefore, we present SynHand5M: a million-scale synthetic dataset with accurate joint annotations, segmentation masks and mesh files of depth maps. Among model based learning (hybrid) methods, we show improved results on our dataset and two of the public benchmarks i.e. NYU and ICVL. Also, by employing a joint training strategy with real and synthetic data, we recover 3D hand mesh and pose from real images in 3.7ms.
关键词： 3D hand pose estimation,hand mesh reconstruction,synthetic dataset,deep learning,CNN
更新于2025-09-04 15:30:14
-
[IEEE 2018 International Conference on 3D Vision (3DV) - Verona (2018.9.5-2018.9.8)] 2018 International Conference on 3D Vision (3DV) - SegmentedFusion: 3D Human Body Reconstruction Using Stitched Bounding Boxes
摘要： This paper presents SegmentedFusion, a method possessing the capability of reconstructing non-rigid 3D models of a human body by using a single depth camera with skeleton information. Our method estimates a dense volumetric 6D motion field that warps the integrated model into the live frame by segmenting a human body into different parts and building a canonical space for each part. The key feature of this work is that a deformed and connected canonical volume for each part is created, and it is used to integrate data. The dense volumetric warp field of one volume is represented efficiently by blending a few rigid transformations. Overall, SegmentedFusion is able to scan a non-rigidly deformed human surface as well as to estimate the dense motion field by using a consumer-grade depth camera. The experimental results demonstrate that SegmentedFusion is robust against fast inter-frame motion and topological changes. Since our method does not require prior assumption, SegmentedFusion can be applied to a wide range of human motions.
关键词： non-rigid,3D reconstruction,skeleton information,depth camera,human body
更新于2025-09-04 15:30:14
-
[IEEE 2018 37th Chinese Control Conference (CCC) - Wuhan (2018.7.25-2018.7.27)] 2018 37th Chinese Control Conference (CCC) - A Lidar-Based Tree Canopy Detection System Development
摘要： In this paper, a LiDAR-based interactive target detection system was developed to characterized tree canopy structures under various laser sensor travel speed and detection distances. The system composed of a sliding motion control system and a Lidar-based target detection unit. The target detection unit used a 2700 range laser scanning sensor to measure target object surface distances based on the time-of-flight principle. The laser sensor travel speed and travel distance was controlled via the control system by specifying a position and speed as a set point. A real-time data acquisition and data post-processing programs were developed based on C++ and MATLAB programming languages respectively. The entire system was tested in the laboratory for a wide range of parameters and operating conditions. The test result showed that the system could detect and characterize tree canopy structure at very low travel speed (0.3 m/s) and high travel speed (5.0 m/s), respectively with acceptable accuracy.
关键词： Tree canopy,three-dimensional image reconstruction,Precision agriculture,LiDAR,Servo system
更新于2025-09-04 15:30:14
-
[IEEE NAECON 2018 - IEEE National Aerospace and Electronics Conference - Dayton, OH, USA (2018.7.23-2018.7.26)] NAECON 2018 - IEEE National Aerospace and Electronics Conference - Real-time 3D scene reconstruction and localization with surface optimization
摘要： A real-time 3D scene reconstruction and localization system with surface optimization is proposed. The dense 3D point cloud model is created by utilizing rotation and orientation invariant feature matching along with loop-closure detection algorithm on RGB-D images in a mobile robot. The high resolution and smooth mesh model is implemented on a GPU based computer through wireless communication.
关键词： localization,RGB-D images,surface optimization,mobile robot,GPU,3D scene reconstruction
更新于2025-09-04 15:30:14
-
Forward projection matrix derivation through Monte-Carlo ray-tracing of KSTAR infra-red imaging video bolometer (IRVB)
摘要： The infrared imaging video bolometer (IRVB) as a foil bolometry technique can be an alternative solution to the conventional resistive bolometer due to its electro-magnetic immunity and 2D plasma profiles. The plasma profile of the IRVB cannot be directly derived from the foil images due to the difference between the foil pixel number and the plasma pixel number and the line integrated nature of the incident rays on the foil. So, it needs tomography such as the Phillips-Thikhonov algorithm. The projection matrix constructing the foil image from the plasma very directly influences the tomography. So, the projection matrix needs to be constructed precisely. For the technique calculating the precise projection matrix, a forward Monte-Carlo ray-tracing method is proposed here, and this technique can provide the detailed descriptions of the foil image. And it can give enhanced performance in the reconstructions of the plasmas with spatially localized power.
关键词： infrared imaging video bolometer,projection matrix,IRVB,foil bolometry,Monte-Carlo ray-tracing,plasma profile reconstruction
更新于2025-09-04 15:30:14