人工智能培训

搜索

人工智能论文:将人类置于场景中:在3D室内环境中学习经济效益(Putting Humans in a Scene: Learning Affordance in

[复制链接]
kjdshfjsfgsfg 发表于 2019-3-15 12:38:06 | 显示全部楼层 |阅读模式
kjdshfjsfgsfg 2019-3-15 12:38:06 252 0 显示全部楼层
人工智能论文:将人类置于场景中:在3D室内环境中学习经济效益(Putting Humans in a Scene: Learning Affordance in 3D Indoor Environments)负担建模在视觉理解中起着重要作用。在本文中,我们的目标是预测3D室内场景的可供性,特别是在给定的室内环境下提供的人体姿势,例如坐在马甲上或站在地板上。为了预测有效的能量和在室内场景中学习可能的3D人体姿势,我们需要了解场景的语义和几何结构以及它与人类的潜在交互。为了学习这样的模型,需要一个3D室内可供性的大型数据集。在这项工作中,我们构建了一个全自动3D姿势合成器,它从电视剧中提取的大量2D姿势以及来自室内体素的体素表示的3D几何知识中融合了语义知识。利用合成器创建的数据,我们引入了一个3D姿势生成模型来预测给定场景中的语义合理和物理上可行的人体姿势(作为单个RGB,RGB-D或深度图像提供)。我们证明了我们的人类可供性预测方法始终如一地完善了现有的最先进方法。
Affordance modeling plays an important role in visual understanding.In thispaper, we aim to predict affordances of 3D indoor scenes, specifically whathuman poses are afforded by a given indoor environment, such as sitting on achair or standing on the floor.In order to predict valid affordances and learnpossible 3D human poses in indoor scenes, we need to understand the semanticand geometric structure of a scene as well as its potential interactions with ahuman.To learn such a model, a large-scale dataset of 3D indoor affordances isrequired.In this work, we build a fully automatic 3D pose synthesizer thatfuses semantic knowledge from a large number of 2D poses extracted from TVshows as well as 3D geometric knowledge from voxel representations of indoorscenes.With the data created by the synthesizer, we introduce a 3D posegenerative model to predict semantically plausible and physically feasiblehuman poses within a given scene (provided as a single RGB, RGB-D, or depthimage).We demonstrate that our human affordance prediction method consistentlyoutperforms existing state-of-the-art methods.人工智能论文:将人类置于场景中:在3D室内环境中学习经济效益(Putting Humans in a Scene: Learning Affordance in 3D Indoor Environments) yVI3LZ7v6Ms2sr28.jpg
URL地址:https://arxiv.org/abs/1903.05690     ----pdf下载地址:https://arxiv.org/pdf/1903.05690    ----人工智能论文:将人类置于场景中:在3D室内环境中学习经济效益(Putting Humans in a Scene: Learning Affordance in 3D Indoor Environments)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

kjdshfjsfgsfg当前离线
新手上路

查看:252 | 回复:0

快速回复 返回顶部 返回列表