RI: Medium: Collaborative Research: Physically Grounded Object Recognition

提案标题：RI：媒介：合作研究：物理接地的物体认知机构：卡内基梅隆大学摘要日期：05/05/09“该奖项是根据2009年美国复苏和再投资法案（公法111-5）资助的。”虽然世界是非常三维的，但今天的大多数视觉对象识别方法基本上都将问题简化为二维模式分类之一，其中矩形图像补丁与存储模板独立比较，以在图像中产生孤立的对象标签。该项目旨在通过探索场景组件之间的三维空间关系、类别级对象模型和全局场景理解方面的定性计量推理，来解释现实世界的三维本质。该项目围绕两个主要研究领域组织。定性3D场景解析：我们努力的核心部分将是开发描述所描绘的对象和表面及其物理关系的场景的定性3D模型。场景中的地面物体：我们将场景的几何表示和相应的三维空间关系与物体识别过程相结合，方法是：(1)根据场景组件之间的三维关系推断出可能的物体身份集；(2)根据场景布局预测最可能出现的目标位置；(3)利用遮挡关系和深度排序来预测场景中可能可见的物体部分。该项目预计将在从照片中理解3D场景方面取得重大进展，这是一项关键的使能技术，可用于广泛的应用，包括自主系统、医疗保健、人机交互、辅助技术、图像检索、工业和个人机器人、制造、科学图像分析、监视和安全以及运输。国家科学基金会提案摘要提案：0905402 PI名称：Hebert， martial印刷自eJacket: 05/06/09第1页1

Proposal Title: RI: Medium: Collaborative Research: Physically Grounded ObjectRecognitionInstitution: Carnegie-Mellon UniversityAbstract Date: 05/05/09"This award is funded under the American Recovery and Reinvestment Act of 2009(Public Law 111-5)."Although the world is very much three-dimensional, most of today's approaches tovisual object recognition essentially reduce the problem to one of 2D patternclassification, where rectangular image patches are independently compared to storedtemplates to produce isolated object labels within the image. This project aims toaccount for the three-dimensional nature of the real world by exploring qualitativegeometric reasoning in terms of 3D spatial relationships between scene components,category-level object models, and global scene understanding.The project is organized around two major research areas. Qualitative 3D sceneparsing: A central part of our effort will be to develop qualitative 3D models of the scenethat describe the depicted objects and surfaces and their physical relations. Groundingobjects in the scene: We integrate the geometric representation of the scene and thecorresponding 3D spatial relations with the object recognition process by (1) inferringthe set of likely object identities based on 3D relations among scene components; (2)predicting the most likely object locations from the scene layout; and (3) using theocclusion relations and depth ordering to predict the parts of objects that may be visiblein the scene.The project is anticipated to result in major advances in 3D scene understanding fromphotographs, a critical enabling technology for a wide range of applications includingautonomous systems, health care, human-computer interaction, assistive technology,image retrieval, industrial and personal robotics, manufacturing, scientific imageanalysis, surveillance and security, and transportation.NATIONAL SCIENCE FOUNDATIONProposal AbstractProposal:0905402 PI Name:Hebert, MartialPrinted from eJacket: 05/06/09 Page 1 of 1