Showing papers in "Computers & Graphics in 2019"
••
TL;DR: Zhang et al. as mentioned in this paper use the soft-argmax function to convert feature maps directly to body joint coordinates, resulting in a fully differentiable framework, which is able to learn heat maps representations indirectly, without additional steps of artificial ground truth generation.
160 citations
••
TL;DR: An improved, targeted sampling strategy as well as an efficient hypothesis validation approach to drastically improve the overall runtime are introduced and several experiments demonstrate the robustness of the proposed approach.
36 citations
••
TL;DR: Positive evidence that agent embodiment induces a higher sense of spatial and social presence is found and both embodied and thematically related audio-visual representations of VAs positively affect the overall user experience.
31 citations
••
TL;DR: Compared with the state-of-the-art methods, the proposed real-time portrait segmentation model, called PortraitNet, achieves remarkable performance in terms of both accuracy and efficiency, especially for generating results with sharper boundaries and under severe illumination conditions.
29 citations
••
TL;DR: This work improves the pedestrian handling systems by incorporating an object tracker with the aim of predicting the pedestrian’s behavior, and is safer and more efficient than the system without tracking functionality due to the early decision capability.
25 citations
••
TL;DR: A novel 3D skeleton human motion data refinement method based on a bidirectional recurrent autoencoder that can exploit the spatial and temporal relationships between previous and subsequent motion data that outperforms several state-of-the-art methods.
24 citations
••
TL;DR: A surface reconstruction algorithm that takes an unoriented point cloud as input and produces an interpolating surface in the form of triangulation that compares well to popular classic and current algorithms with unoriented input points and triangulated surface output.
23 citations
••
TL;DR: An Anatomy Studio, a collaborative Mixed Reality tool for virtual dissection that combines tablets with styli and see-through head-mounted displays to assist anatomists by easing manual tracing and exploring cryosection images, and contributes novel interaction techniques intended to promote spatial understanding and expedite manual segmentation.
22 citations
••
TL;DR: In this paper, the authors propose a visually realistic, flexible and robust grasping system that enables real-time interactions in virtual environments, where a hand is automatically fitted to the object shape from a position and orientation determined by the user using the VR handheld controllers (e.g. Oculus Touch motion controllers).
22 citations
••
TL;DR: The software Bloodline is presented, a tool combining advanced visual exploration and analysis of 4D PC-MRI data that provides a guided data processing and evaluation workflow and focuses on the detection and characterization of vortices, which are correlated to cardiovascular diseases.
19 citations
••
TL;DR: An end-to-end pipeline capable of finding the size of the minimal repeating pattern in single images, as well as obtaining the single repetition that, when tiled, produces the most similar synthesis to the complete image is proposed.
••
TL;DR: This work introduces a crowd simulation method that runs at interactive rates for on the order of a hundred thousand agents, making it particularly suitable for use in games.
••
TL;DR: Yarn is extended, a system for automatic construction and visualization of narratives with multiple timelines, and constructs both linear as well as non-linear narratives, and enables pairwise comparison within a group of multiple narrative timelines.
••
TL;DR: A method to derive a robust skeleton representation from a depth sequence and to complement it with a highly discriminative face feature to improve rank-1 re-identification accuracy compared to individual cues is proposed.
••
TL;DR: This work develops the theoretical and computational foundations leading to a generalized procedure for generating Delaunay Lofts — a new class of scutoid-like shapes inspired by “scutoids” — shapes that were recently reported to occur in epithelial cells due to topological changes between the extremal surfaces of epithelia.
••
TL;DR: An efficient and practically robust method to isotropically remesh an input triangular mesh with bounded approximation error based on a key observation that is, when more uniformly distributed vertices are added into the remeshed mesh, the error-bounded constraint is usually satisfied.
••
TL;DR: Environmental factors can provide a reasonably generalizable approach to increase social presence in HMD-based MR environments beyond the specific scenario and environment described here.
••
TL;DR: Different neural network architectures and local neighborhood representations that improve sharpness field computation are extended, resulting in significantly improved feature-aware smoothing and the accuracy of different deep learning models for sharpity field computation is evaluated.
••
TL;DR: A deep learning framework for unsupervised 3D shape retrieval with geodesic moments is presented, showing that it leads to improved retrieval results in comparison with state-of-the-art techniques.
••
TL;DR: A non-parametric (and hence threshold independent) algorithm for segmenting CAD models (represented as meshes) which does not require any user intervention and comparison with the state-of-the-art approaches indicate that it performs better or on par.
••
TL;DR: A quantitative user study has conducted evaluating the relative strengths and weaknesses of each interface in relation to different tasks in immersive environments, showing significant differences on the interface performances according to different parameters.
••
TL;DR: This work is the first to explore the domain of LSHF CAR and provides insight into designing experiences in other AR domains and creates HoloRoyale, the first large scale high fidelity (LSHF) CAR experience.
••
TL;DR: BrainTrawler is proposed, a task-driven, web-based framework that incorporates visual analytics methods to explore heterogeneous neurobiological data and facilitates spatial indexing to query large-scale voxel-level connectivity data and gene expression collections in real-time.
••
TL;DR: A robust sample-based GPU algorithm is presented for calculating the full topology of Voronoi diagrams of non-general position spheres, and efficiently and robustly handles all input, whether in general or non- general position, and finds all Vor onoi vertices and edges, modulo the sampling density, including isolated disconnected edges.
••
TL;DR: This research proposed total absolute curvature to quantify “complexity“, which is one of the shape features related to “beauty”, and described its geometric meaning as the total area of the Gauss map on the curved surface shape.
••
TL;DR: It is verified that color normalization does not contribute to the classification of breast tumors when associated with percolation features, however, color normalized images from the colorectal tumor’s dataset provided better results than the original images.
••
TL;DR: This work proposes an interactive dual-visual approach whereby changes and selections in attribute space reflect visually on the configuration of data layouts, conceived to support immediate analysis of the impact of selected subsets of attributes in the organization of the data set.
••
TL;DR: A novel method for carving a QR code on a surface in a perforated manner and fabricating it with single material via ordinary 3D printing techniques, and introduces a QR evaluator to estimate its readability and then optimize the code in terms of the printability and robustness.
••
TL;DR: A new tool to sculpt the geometry of the environment and the molecules is added and a new history based visualization is introduced that enables the user to interactively explore and distil, from the busy and chaotic mesoscale environment, the interactions between molecules that drive cellular processes.
••
TL;DR: A novel slice-based physical model is proposed as a complementary method for visualizing blood flow data and it is demonstrated that the technique is capable of transforming volumetric flow data into a tangible, easily fabricable model.