RI: Medium: Creating Knowledge with All-Novel-Class Computer Vision
RI:媒介:利用新颖的计算机视觉创造知识
基本信息
- 批准号:2106825
- 负责人:
- 金额:$ 120万
- 依托单位:
- 依托单位国家:美国
- 项目类别:Continuing Grant
- 财政年份:2021
- 资助国家:美国
- 起止时间:2021-09-01 至 2025-08-31
- 项目状态:未结题
- 来源:
- 关键词:
项目摘要
Computer vision methods have had very high impact on science and industry, but this impact has been confined to cases where there is access to very large quantities of labelled data (i.e., objects are identified in the image), which have either been published, collected or purchased. This research will study computer vision methods that operate in areas where there are very little labelled data. This project builds on the natural model of the way humans and animals learn to label images. One core research goal is an object detection procedure that can be trained with all category data — there will be a small number of examples each from a large number of categories. Another core goal is a learning procedure that can share training examples across categories widely and effectively without explicit linking of the categories. A third core goal is linking learning of early vision tasks — for example, recovering shading and lighting from an image — to learning of classification and detection tasks, so that both tasks can be learned with very little labelled data. Successful completion of this research will unify apparently disparate areas of computer vision, by linking early vision and categorization directly, and will create novel methods for improving categorization performance in difficult circumstances. Furthermore, successful completion of this research will unlock many real-world applications that need all category methods. The all-novel-class problem occurs where there are a small number of examples each from a large number of classes and no class has many examples. This project addresses the all-novel-class issue by sharing of various kinds of information during training. Specifically, three kinds of sharing principles will be studied. The first is a cell consistency principle that uses a geometric and probabilistic analysis of class boundaries driven by feature generation to produce improvements in classification, by requiring that the cells in feature space associated with the similar classes. Second, a label consistency principle uses probabilistic reasoning to impute labels and confidence weights for unlabeled examples. Label consistency can be applied extensively, from category labels for individual examples to superclass labels that identify classes over which sharing will be helpful. Finally, a physical consistency principle requires that inferences from images are consistent with simple physics laws; this principle allows researchers to impute missing annotations for early vision data and links high level classes to early vision through an attribute theory.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
计算机视觉方法已经对科学和工业产生了非常高的影响,但这种影响仅限于可以访问非常大量的标记数据的情况(即,在图像中识别对象),这些对象或者已经出版、被收集或者被购买。这项研究将研究在标记数据很少的领域中运行的计算机视觉方法。这个项目建立在人类和动物学习标记图像的自然模型上。一个核心研究目标是一个可以用所有类别数据训练的对象检测过程-每个类别都有少量的例子。另一个核心目标是一个学习过程,可以广泛有效地跨类别共享训练示例,而无需明确链接类别。第三个核心目标是将早期视觉任务的学习(例如,从图像中恢复阴影和照明)与分类和检测任务的学习联系起来,这样两个任务都可以用很少的标记数据来学习。这项研究的成功完成将通过将早期视觉和分类直接联系起来,统一计算机视觉的明显不同的领域,并将创造新的方法来改善困难情况下的分类性能。此外,这项研究的成功完成将解锁许多需要所有类别方法的实际应用。全新类问题发生在有少量的例子,每个例子来自大量的类,没有类有很多例子。该项目通过在培训过程中共享各种信息来解决全小说类问题。具体而言,将研究三种共享原则。第一个是细胞一致性原则,它使用由特征生成驱动的类边界的几何和概率分析来产生分类的改进,通过要求特征空间中的细胞与相似的类相关联。其次,标签一致性原则使用概率推理来估算未标记示例的标签和置信度权重。标签一致性可以广泛应用,从单个示例的类别标签到标识类的超类标签,在这些类上共享将是有帮助的。最后,一个物理一致性原则要求从图像的推论是符合简单的物理定律;这一原则允许研究人员归咎于早期视力数据的注释丢失,并通过属性理论链接到早期视力的高级类。该奖项反映了NSF的法定使命,并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。
项目成果
期刊论文数量(51)
专著数量(0)
科研奖励数量(0)
会议论文数量(0)
专利数量(0)
CEIP: Combining Explicit and Implicit Priors for Reinforcement Learning with Demonstrations
- DOI:10.48550/arxiv.2210.09496
- 发表时间:2022-10
- 期刊:
- 影响因子:0
- 作者:Kai Yan;A. Schwing;Yu-Xiong Wang
- 通讯作者:Kai Yan;A. Schwing;Yu-Xiong Wang
SDFusion: Multimodal 3D Shape Completion, Reconstruction, and Generation; CVPR; 2023
SDFusion:多模态 3D 形状完成、重建和生成;
- DOI:
- 发表时间:2023
- 期刊:
- 影响因子:0
- 作者:Y.-C. Cheng;H.-Y. Lee;S. Tulyakov;A.G. Schwing;L. Gui
- 通讯作者:L. Gui
Learnable Polyphase Sampling for Shift Invariant and Equivariant Convolutional Networks
- DOI:10.48550/arxiv.2210.08001
- 发表时间:2022-10
- 期刊:
- 影响因子:0
- 作者:Renan A. Rojas-Gomez;Teck-Yian Lim;A. Schwing;M. Do;Raymond A. Yeh
- 通讯作者:Renan A. Rojas-Gomez;Teck-Yian Lim;A. Schwing;M. Do;Raymond A. Yeh
Is Self-Supervised Learning More Robust Than Supervised Learning?
- DOI:10.48550/arxiv.2206.05259
- 发表时间:2022-06
- 期刊:
- 影响因子:0
- 作者:Yuanyi Zhong;Haoran Tang;Jun-Kun Chen;Jian Peng;Yu-Xiong Wang
- 通讯作者:Yuanyi Zhong;Haoran Tang;Jun-Kun Chen;Jian Peng;Yu-Xiong Wang
Diverse Human Motion Prediction Guided by Multi-level Spatial-Temporal Anchors
- DOI:10.1007/978-3-031-20047-2_15
- 发表时间:2023-02
- 期刊:
- 影响因子:0
- 作者:Sirui Xu;Yu-Xiong Wang;Liangyan Gui
- 通讯作者:Sirui Xu;Yu-Xiong Wang;Liangyan Gui
{{
item.title }}
{{ item.translation_title }}
- DOI:
{{ item.doi }} - 发表时间:
{{ item.publish_year }} - 期刊:
- 影响因子:{{ item.factor }}
- 作者:
{{ item.authors }} - 通讯作者:
{{ item.author }}
数据更新时间:{{ journalArticles.updateTime }}
{{ item.title }}
- 作者:
{{ item.author }}
数据更新时间:{{ monograph.updateTime }}
{{ item.title }}
- 作者:
{{ item.author }}
数据更新时间:{{ sciAawards.updateTime }}
{{ item.title }}
- 作者:
{{ item.author }}
数据更新时间:{{ conferencePapers.updateTime }}
{{ item.title }}
- 作者:
{{ item.author }}
数据更新时间:{{ patent.updateTime }}
David Forsyth其他文献
Supplement - Convex Decomposition of Indoor Scenes
- DOI:
- 发表时间:
- 期刊:
- 影响因子:0
- 作者:
David Forsyth - 通讯作者:
David Forsyth
Hidden Markov Models
隐马尔可夫模型
- DOI:
10.1007/978-3-030-18114-7_13 - 发表时间:
2019 - 期刊:
- 影响因子:0
- 作者:
David Forsyth - 通讯作者:
David Forsyth
Preserving Image Properties Through Initializations in Diffusion Models
通过扩散模型中的初始化保留图像属性
- DOI:
10.1109/wacv57701.2024.00516 - 发表时间:
2024 - 期刊:
- 影响因子:0
- 作者:
Jeffrey Zhang;Shao;Kedan Li;David Forsyth - 通讯作者:
David Forsyth
Fully spectrum-sliced four-wave mixing wavelength conversion in a Semiconductor Optical Amplifier
半导体光放大器中的全光谱切片四波混频波长转换
- DOI:
- 发表时间:
2024 - 期刊:
- 影响因子:0.5
- 作者:
David Forsyth - 通讯作者:
David Forsyth
Scientific report on Modeling and Prediction of Human Intent for Primitive Activation
关于人类原始激活意图的建模和预测的科学报告
- DOI:
- 发表时间:
2014 - 期刊:
- 影响因子:0
- 作者:
David Forsyth - 通讯作者:
David Forsyth
David Forsyth的其他文献
{{
item.title }}
{{ item.translation_title }}
- DOI:
{{ item.doi }} - 发表时间:
{{ item.publish_year }} - 期刊:
- 影响因子:{{ item.factor }}
- 作者:
{{ item.authors }} - 通讯作者:
{{ item.author }}
{{ truncateString('David Forsyth', 18)}}的其他基金
Collaborative Research: Computational Behavioral Science: Modeling, Analysis, and Visualization of Social and Communicative Behavior
合作研究:计算行为科学:社交和交流行为的建模、分析和可视化
- 批准号:
1029035 - 财政年份:2010
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
RI: Small: Exploiting Geometric and Illumination Context in Indoor Scenes
RI:小:利用室内场景中的几何和照明环境
- 批准号:
0916014 - 财政年份:2009
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
INT2-Medium: Understanding the meaning of images
INT2-Medium:理解图像的含义
- 批准号:
0803603 - 财政年份:2008
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
Interpreting Human Behaviour in Video using FSA's and Object Context
使用 FSA 和对象上下文解释视频中的人类行为
- 批准号:
0534837 - 财政年份:2006
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
Finding and Tracking People from the Bottom Up
自下而上查找和跟踪人员
- 批准号:
0098682 - 财政年份:2001
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
SGER: MCMC Algorithms for Object Recognition
SGER:用于对象识别的 MCMC 算法
- 批准号:
9979201 - 财政年份:1999
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
A Spiral Approach to Chemical Concepts Using GC/MS
使用 GC/MS 探索化学概念的螺旋方法
- 批准号:
9850580 - 财政年份:1998
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
Workshop on Shape, Contour and Grouping
形状、轮廓和分组研讨会
- 批准号:
9712426 - 财政年份:1997
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
Recognising curved surfaces from their outlines
从轮廓识别曲面
- 批准号:
9596025 - 财政年份:1994
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
相似海外基金
DESIGN: Creating cultural change in small to medium-sized professional societies: a training network approach
设计:在中小型专业团体中创造文化变革:培训网络方法
- 批准号:
2334964 - 财政年份:2024
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
CCRI: Medium: MSP-Podcast: Creating The Largest Speech Emotional Database By Leveraging Existing Naturalistic Recordings
CCRI:媒介:MSP-Podcast:利用现有的自然主义录音创建最大的语音情感数据库
- 批准号:
2016719 - 财政年份:2020
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
Development of basic technology for creating mucosal vaccines using unique medium-chain fatty acids secreted by worker bees
开发利用工蜂分泌的独特中链脂肪酸制造粘膜疫苗的基础技术
- 批准号:
19H03384 - 财政年份:2019
- 资助金额:
$ 120万 - 项目类别:
Grant-in-Aid for Scientific Research (B)
Creating a Chronotopic Ground for the Mapping of Literary Texts: Innovative Data Visualisation and Spatial Interpretation in the Digital Medium
为文学文本的映射创造时间主题基础:数字媒体中的创新数据可视化和空间解释
- 批准号:
AH/P00895X/1 - 财政年份:2017
- 资助金额:
$ 120万 - 项目类别:
Research Grant
CIF: Medium: Collaborative Research: Interference-Aware Cooperation via Structured Codes: Creating an Empirical Cycle
CIF:媒介:协作研究:通过结构化代码进行干扰感知合作:创建经验循环
- 批准号:
1302630 - 财政年份:2013
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
SBIR Phase I: Non-linear Mechanical System for Creating Propulsive Thrust in a Fluid Medium
SBIR 第一阶段:在流体介质中产生推进推力的非线性机械系统
- 批准号:
1315383 - 财政年份:2013
- 资助金额:
$ 120万 - 项目类别:
Standard Grant
CIF: Medium: Collaborative Research: Interference-Aware Cooperation via Structured Codes: Creating an Empirical Cycle
CIF:媒介:协作研究:通过结构化代码进行干扰感知合作:创建经验循环
- 批准号:
1302600 - 财政年份:2013
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
CIF: Medium: Collaborative Research: Interference-Aware Cooperation via Structured Codes: Creating an Empirical Cycle
CIF:媒介:协作研究:通过结构化代码进行干扰感知合作:创建经验循环
- 批准号:
1302616 - 财政年份:2013
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
RI: Medium: Collaborative Research: Creating Organizationally Adept Software Agents and their Organizations
RI:媒介:协作研究:创建组织熟练的软件代理及其组织
- 批准号:
0964590 - 财政年份:2010
- 资助金额:
$ 120万 - 项目类别:
Continuing Grant
SHF: Medium: Assurance Based Development: A Rational Approach To Creating High Assurance Software
SHF:中:基于保证的开发:创建高保证软件的合理方法
- 批准号:
0905375 - 财政年份:2009
- 资助金额:
$ 120万 - 项目类别:
Standard Grant