EAGER: Improving the Quality and Editability of 2D and 3D Shapes via Crowdsourcing and Self-Crowdsourcing

二维曲线和三维曲面是计算机图形学的核心，但目前创建和编辑它们的方法受到用户专业知识的限制。这些任务中的许多都可以由任何人在基本水平上完成，但很难达到专家水平。PI在这项探索性研究中的目标是寻求超越基于众包的这种限制的方法，在众包中，群体超越个人的技能（通过许多个人贡献单一输入），个人超越自己的能力（通过贡献许多输入），这可能被称为自我众包（PI指出，众包的改进效果甚至适用于一个人的群体）。项目成果不仅会对计算机辅助设计的无数应用产生积极影响，还会对计算机辅助手术等领域产生积极影响，在这些领域，切割路径的质量可能具有生死攸关的重要性。患有各种运动障碍的人也应该受益于在诸如签署文件之类的日常任务中实现高质量身体表现的能力。为了确保他的发现得到广泛传播，PI将开放作为这项研究的一部分开发的软件的源代码。PI的方法是跨学科的，融合了人机交互、实验心理学和计算机图形学。人群的智慧已经被研究用于估计单个数字的场景，比如一件物品的重量或价格；然而，平均（或以其他方式聚合）高维物理或创造性输入的技术尚未得到研究。这项研究将涉及三个互补的重点：分而治之的方法，用于结合来自一群（新手）个人的小贡献，从2D图像中创建高质量的3D形状；将图像和形状创建的时间记录保存在称为时间通道的可重用表示中，以提高可编辑性的算法；以及用于聚合重复2D和3D曲线绘图的算法（自众包）。这项工作将为理解众包的局限性奠定基础；特别是，众包的2D-to-3D算法将测试几个长期存在的关于新手创建3D形状的假设。

Two-dimensional curves and three-dimensional surfaces lie at the core of computer graphics, but current approaches to creating and editing them are limited by the user's expertise. Many of these tasks can be performed by anyone at a basic level, but expert performance is difficult to achieve. The PI's goal in this exploratory research is to seek ways to transcend this limitation based on crowdsourcing, in which a crowd surpasses the skill of the individual (by many individuals each contributing a single input), and an individual surpasses his or her own abilities (by contributing many inputs) in what might be termed self-crowdsourcing (the PI points out it has been shown that the improving effects of crowdsourcing apply even to a crowd of one). Project outcomes will positively impact not only the myriad applications of computer-aided design but also fields such as computer-assisted surgery, where the quality of a cutting path can take on life-or-death importance. Individuals with a variety of motor impairments should also benefit from the ability to achieve high-quality physical performance in tasks as routine as signing a document. To ensure broad dissemination of his findings, the PI will open source the software developed as part of this research.The PI's approach is interdisciplinary, fusing human-computer interaction, experimental psychology, and computer graphics. The wisdom of crowds has been studied for scenarios where a single number is estimated, such as the weight or price of an item; however, techniques for averaging (or otherwise aggregating) high-dimensional physical or creative inputs have not been studied. This research will involve three complementary thrusts: divide-and-conquer approaches for combining small contributions from a crowd of (novice) individuals to create a high-quality 3D shape from a 2D image; algorithms for preserving the temporal record of image and shape creation in a reusable representation called a time channel, to improve editability; and algorithms for aggregating repeated 2D and 3D curve drawings (self-crowdsourcing). The work will lay the foundation for an understanding of the limits of crowdsourcing; in particular, the crowdsourced 2D-to-3D algorithm will test several long-standing hypotheses relating to novice creation of 3D shapes.