A5077188936- Advanced Vision and Imaging
- Computer Graphics and Visualization Techniques
- Image Enhancement Techniques
- Ethics and bioethics in healthcare
- 3D Shape Modeling and Analysis
- Philosophical Thought and Analysis
- Topic Modeling
- Generative Adversarial Networks and Image Synthesis
- Natural Language Processing Techniques
- Religious and Theological Studies
- Advanced Image and Video Retrieval Techniques
- Social Sciences and Policies
- Multimodal Machine Learning Applications
- Advanced Text Analysis Techniques
- Advanced Image Processing Techniques
- Economic and Social Development
- Theology and Canon Law Studies
- Video Analysis and Summarization
- Historical and Literary Analyses
- Medical Image Segmentation Techniques
- Sentiment Analysis and Opinion Mining
- Literary and Cultural Studies
- Color Science and Applications
- Image Retrieval and Classification Techniques
- Spanish Philosophy and Literature
Adobe Systems (United States)
2017-2025
Universidad de Zaragoza
1970-2024
Instituto de Investigación Sanitaria Aragón
2020
Walt Disney (United States)
2014
We introduce an extremely scalable and efficient yet simple palette-based image decomposition algorithm. Given RGB set of palette colors, our algorithm decomposes the into a additive mixing layers, each which corresponds to color applied with varying weight. Our approach is based on geometry images in RGBXY-space. This new geometric orders magnitude more than previous work requires no numerical optimization. provide implementation 48 lines Python code. demonstrate real-time layer tool users...
We present a palette-based framework for color composition visual applications and three large-scale, wide-ranging perceptual studies on the perception of harmonization. abstract relationships between palette colors as compact set axes describing harmonic templates over perceptually uniform wheels. Our provides basis interactive color-aware operations such harmonization images videos. Because our approach to is palette-based, we are able conduct first controlled experiments evaluating...
Recently, we have seen a growing trend in the design and fabrication of personalized figurines, created by scanning real people then physically reproducing miniature statues with 3D printers. This is currently hot topic both academia industry, printed figurines are gaining more realism, especially state-of-the-art facial technology improving. However, current systems all contain same limitation - no previous method able to suitably capture hair-styles for physical reproduction. Typically,...
Amirreza Shirani, Franck Dernoncourt, Paul Asente, Nedim Lipka, Seokhwan Kim, Jose Echevarria, Thamar Solorio. Proceedings of the 57th Annual Meeting Association for Computational Linguistics. 2019.
We present an interactive approach to synthesizing realistic variations in facial hair images, ranging from subtle edits existing the addition of complex and challenging images clean-shaven subjects. To circumvent tedious computationally expensive tasks modeling, rendering compositing 3D geometry target hairstyle using traditional graphics pipeline, we employ a neural network pipeline that synthesizes detailed directly image under one second. The synthesis is controlled by simple sparse...
Compound icons are prevalent on signs, webpages, and infographics, effectively conveying complex abstract concepts, such as "no smoking" "health insurance", with simple graphical representations. However, designing requires experience creativity, in order to efficiently navigate the semantics, space, style features of icons. In this paper, we aim automate process generating given compound facilitate rapid icon creation ideation. Informed by ethnographic interviews professional designers,...
Abstract We present a new image‐based, post‐processing antialiasing technique, which offers practical solutions to the common, open problems of existing filter‐based real‐time algorithms. Some features include local contrast analysis for more reliable edge detection, and simple effective way handle sharp geometric diagonal lines. This, along with our accelerated accurate pattern classification allows better reconstruction silhouettes. Our method shows first time how combine morphological...
We present a method that generates expressive talking-head videos from single facial image with audio as the only input. In contrast to previous attempts learn direct mappings raw pixels for creating talking faces, our first disentangles content and speaker information in input signal. The robustly controls motion of lips nearby regions, while determines specifics expressions rest dynamics. Another key component is prediction landmarks reflecting speaker-aware Based on this intermediate...
Media streaming, with an edge-cloud setting, has been adopted for a variety of applications such as entertainment, visualization, and design. Unlike video/audio streaming where the content is usually consumed passively, virtual reality require 3D assets stored on edge to facilitate frequent edge-side interactions object manipulation viewpoint movement. Compared audio video asset often requires larger data sizes yet lower latency ensure sufficient rendering quality, resolution, perceptual...
We develop a method to automatically segment font’s glyphs into set of overlapping and intersecting strokes with the aim generating artistic stylizations. The segmentation relies on geometric analysis glyph’s outline, its interior, surrounding areas is grounded in perceptually informed principles measures. Our does not require training data or templates applies large variety input languages, writing systems, styles. It uses medial axis, curvilinear shape features that specify convex concave...
Abstract We present a novel method to estimate an approximation of the reflectance characteristics optically thick, homogeneous translucent materials using only single photograph as input. First, we approximate diffusion profile linear combination piecewise constant functions, approach that enables system minimization and maximizes robustness in presence suboptimal input data inferred from image. then fit smoother monotonically decreasing model, ensuring continuity on its first derivative....
In this paper, we present the main findings and compare results of SemEval-2020 Task 10, Emphasis Selection for Written Text in Visual Media. The goal shared task is to design automatic methods emphasis selection, i.e. choosing candidates textual content enable automated assistance authoring. focus on short text instances social media, with a variety examples, from media posts inspirational quotes. Participants were asked model using plain no additional context user or other considerations....
Color editing in images often consists of two main tasks: changing hue and saturation, lightness or tone curves. State-of-the-art palette-based recoloring approaches entangle these tasks. A user's only control is the individual palette colors. This inferior to state-of-the-art commercial software, where based on flexible curves that remap lightness. However, are provided globally per color channel (e.g., RGB). They unrelated image content. Neither nor support direct image-space...
Although color vision deficiencies are very prevalent, there no ideal methods for assessing in all environments. We compared a new digital and automated method that quantifies perception the three protan, deutan, tritan axes with two of most commonly used tests daily practice: Ishihara 38 plates test Farnsworth–Munsell 100-Hue test. One hundred patients underwent triple examination composed DIVE Color Test, test, The Test was performed twice forty participants to assess its repeatability. In...
Photographic composition is often taught as alignment with grids-most commonly, the rule of thirds. Professional photographers use more complex grids, like harmonic armature, to achieve diverse dynamic compositions. We are interested in understanding whether these grids helpful amateurs.
In this paper, we aim to learn associations between visual attributes of fonts and the verbal context texts they are typically applied to. Compared related work leveraging surrounding context, choose focus only on input text, which can enable new applications for text is element in document. We introduce a dataset, containing examples different topics social media posts ads, labeled through crowd-sourcing. Due subjective nature task, multiple might be perceived as acceptable an makes problem...
Introduction Around 70% to 80% of the 19 million visually disabled children in world are due a preventable or curable disease, if detected early enough. Vision screening childhood is an evidence-based and cost-effective way detect visual disorders. However, current programmes face several limitations: training required perform them efficiently, lack accurate tools poor collaboration from young children. Some these limitations can be overcome by new digital tools. Implementing system based on...
We present a method to automatically decompose light field into its intrinsic shading and albedo components. Contrary previous work targeted 2D single images videos, is 4D structure that captures non-integrated incoming radiance over discrete angular domain. This higher dimensionality of the problem renders state-of-the-art algorithms impractical either due their cost processing slice, or inability enforce proper coherence in additional dimensions. propose new decomposition algorithm jointly...
Abstract This paper introduces a new method for simulating homogeneous subsurface light transport in translucent objects. Our approach is based on irradiance convolutions over multi‐layered representation of the volume transport, which general enough to obtain plausible depictions objects diffusion approximation. We aim at providing an efficient physically algorithm that can apply arbitrary profiles geometries. accurate results wide range materials, par with hierarchical by Jensen and Buhler.
Current image harmonization methods consider the entire background as guidance for harmonization. However, this may limit capability user to choose any specific object/person in guide To enable flexible interaction between and harmonization, we introduce interactive a new setting where is performed with respect selected \emph{region} reference instead of background. A framework that allows users pick certain regions use it proposed. Inspired by professional portrait users, also luminance...