Low-Dimensionality Calibration through Local Anisotropic Scaling for Robust Hand Model Personalization

Remelli, Edoardo; Tkach, Anastasia; Tagliasacchi, Andrea; Pauly, Mark

doi:10.1109/iccv.2017.277

Cited by 41 publications

(36 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The studies of hand pose estimation from 3D data have made great progress in recent years, along with the development of depth sensors [34,35]. These studies are mainly based on three models, which are generative models [35][36][37][38][39], discriminative models [5,[10][11][12]40,41] and hybrid models, separately [42][43][44]. Generally, hand pose estimation methods can be divided into three categories: generative methods [35][36][37][38][39], discriminative methods [5,[10][11][12]40,41] and hybrid methods [42][43][44].…”

Section: Hand Pose Estimationmentioning

confidence: 99%

3D Capsule Hand Pose Estimation Network Based on Structural Relationship Information

Zhang

et al. 2020

Symmetry

View full text Add to dashboard Cite

Hand pose estimation from 3D data is a key challenge in computer vision as well as an essential step for human–computer interaction. A lot of deep learning-based hand pose estimation methods have made significant progress but give less consideration to the inner interactions of input data, especially when consuming hand point clouds. Therefore, this paper proposes an end-to-end capsule-based hand pose estimation network (Capsule-HandNet), which processes hand point clouds directly with the consideration of structural relationships among local parts, including symmetry, junction, relative location, etc. Firstly, an encoder is adopted in Capsule-HandNet to extract multi-level features into the latent capsule by dynamic routing. The latent capsule represents the structural relationship information of the hand point cloud explicitly. Then, a decoder recovers a point cloud to fit the input hand point cloud via a latent capsule. This auto-encoder procedure is designed to ensure the effectiveness of the latent capsule. Finally, the hand pose is regressed from the combined feature, which consists of the global feature and the latent capsule. The Capsule-HandNet is evaluated on public hand pose datasets under the metrics of the mean error and the fraction of frames. The mean joint errors of Capsule-HandNet on MSRA and ICVL datasets reach 8.85 mm and 7.49 mm, respectively, and Capsule-HandNet outperforms the state-of-the-art methods on most thresholds under the fraction of frames metric. The experimental results demonstrate the effectiveness of Capsule-HandNet for 3D hand pose estimation.

show abstract

Section: Hand Pose Estimationmentioning

confidence: 99%

3D Capsule Hand Pose Estimation Network Based on Structural Relationship Information

Zhang

et al. 2020

Symmetry

View full text Add to dashboard Cite

show abstract

“…To this end, studies are being conducted on computing joint movements from videos captured with optical markers attached to the surfaces of a subject's joints [16]; as well as on detecting and recognizing the movements of the whole body, beyond merely the hands, with various equipment, including the Kinect and Leap Motion [2,17]. In recent years, studies have been conducted on detection models for computing movements made with elaborate hand joint models [18], and on realistically representing user behavior in a virtual space from video data obtained with motion capture equipment [6]. These studies have been performed to improve the sense of presence by representing user behavior in a virtual environment and enabling users to provide a variety of physical responses as direct feedback, using their hands or other body parts.…”

Section: Immersive Virtual Realitymentioning

confidence: 99%

A Study on the Presence of Immersive User Interface in Collaborative Virtual Environments Application

et al. 2019

View full text Add to dashboard Cite

This study proposes a collaboration-based interaction as a new method for providing an improved presence and a satisfying experience to various head-mounted display (HMD) users utilized in immersive virtual reality (IVR), and analyzes the experiences (improved presence, satisfying enjoyment, and social interaction) of applying collaboration to user interfaces. The key objective of the proposed interaction is to provide an environment where HMD users are able to collaborate with each other, based on their differentiated roles and behaviors. To this end, a collaboration-based interaction structured in three parts was designed, including a synchronization procedure and a communication interface that enable users to swiftly execute common goals with precision, based on immersive interactions that allow users to directly exchange information and provide feedback with their hands and feet. Moreover, experimental VR applications were built to systematically analyze the improved presence, enjoyment, and social interaction experienced by users through collaboration. Finally, by conducting a survey on the participants of the experiment, this study confirmed that the proposed interface indeed provided users with an improved presence and a satisfying experience, based on collaboration.

show abstract

“…By attaching a surface or optical marker to a hand joint and detecting movements based on the marker, many studies including that of Matcalf et al [6] have proposed methods of mapping the detecting movements to the motions of a virtual hand model [7,8]. Recently, techniques that express realistic movements in virtual spaces have been developed, e.g., a sphere-mesh tracking model proposed by Remellli et al [9] and Total Capture (hand gestures, facial expressions, and whole-body movements are captured) proposed by Joo et al [10].…”

Section: Immersive Vrmentioning

confidence: 99%

Maze Terrain Authoring System in Immersive Virtual Reality for New Visual Realism

Kim

2019

Symmetry

View full text Add to dashboard Cite

This paper proposes a maze terrain authoring system with which a user can automatically calculate various and complex maze patterns to compose maze terrains in an easier and more intuitive structure. Using the maze terrain information calculated by using the proposed authoring system, a 3D maze terrain is generated quickly and effectively, and through this, the user’s visual realism in an immersive virtual reality is increased to provide a new presence. The proposed maze terrain authoring system consists of three core functions: a function that automatically generates a grid maze of various sizes and patterns based on a maze generation algorithm; a function that calculates a circular maze in an intuitive structure; and a function that transforms a sketch-based maze by using an image-processing algorithm. Through the authoring system consisting of these functions, various maze terrains from uniform mazes to irregular mazes can be generated effectively. This study produces maze terrains of various concepts directly by using maze information calculated through the authoring system. In addition, through experiments, this study analyzes the efficiency in an immersive virtual reality and the presence through visual realism. In this process, the suitability of the authoring system is also analyzed in combination with a survey.

show abstract

Low-Dimensionality Calibration through Local Anisotropic Scaling for Robust Hand Model Personalization

Cited by 41 publications

References 35 publications

3D Capsule Hand Pose Estimation Network Based on Structural Relationship Information

3D Capsule Hand Pose Estimation Network Based on Structural Relationship Information

A Study on the Presence of Immersive User Interface in Collaborative Virtual Environments Application

Maze Terrain Authoring System in Immersive Virtual Reality for New Visual Realism

Contact Info

Product

Resources

About