CCRI: Planning: Collaborative Research: Planning to Develop a Low-Power Computer Vision Platform to Enhance Research in Computing Systems

CCRI：规划：协作研究：规划开发低功耗计算机视觉平台以加强计算系统研究

• 批准号: 1925713
• 负责人: Yung-Hsiang Lu
• 金额: $ 5.5万
• 依托单位: Purdue University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-10-01 至 2022-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1925713&HistoricalAwards=false
• 关键词: CCRI; Planning; Collaborative; Research; Develop

This project is a planning grant to investigate how to develop a research infrastructure for investigating energy-efficient computer vision. Computer vision is a set of technologies for understanding image or video. These technologies are essential in many applications, such as autonomous vehicles and airport security. Some vision systems need to be energy-efficient because they use batteries, for example, drones and mobile phones. This infrastructure is planned to be available for researchers to experiment and evaluate their solutions. The evaluation includes the success rates of identifying different types of objects, the speed, and the energy consumption. The infrastructure has three major components: (1) video taken by drones, (2) vision system analyzing the data, (3) power meter. It will be designed for a wide range of vision problems. It will also include a referee system which automates the entire evaluation process. There are many beneficial applications for low-power computer vision, for example, lightweight cameras that may provide safety for the vision impaired, wearable cameras that may improve factory safety or perform inventory analysis, and cameras for observing wildlife. The planning grant is to assess the needs of such an infrastructure for academia and industry. This team will actively engage the research community.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.

该项目是一项规划拨款，旨在研究如何开发研究基础设施，以研究节能计算机视觉。计算机视觉是一组用于理解图像或视频的技术。这些技术在许多应用中至关重要，例如自动驾驶汽车和机场安全。一些视觉系统需要节能，因为它们使用电池，例如无人机和移动的电话。该基础设施计划可供研究人员实验和评估他们的解决方案。评估包括识别不同类型对象的成功率、速度和能量消耗。该基础设施有三个主要组成部分：（1）无人机拍摄的视频，（2）分析数据的视觉系统，（3）功率计。 它将被设计用于广泛的视力问题。 它还将包括一个裁判系统，使整个评估过程自动化。 低功耗计算机视觉有许多有益的应用，例如，可以为视力受损者提供安全的轻型相机，可以提高工厂安全性或执行库存分析的可穿戴相机，以及用于观察野生动物的相机。规划补助金是为了评估学术界和工业界对这种基础设施的需求。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Low-Power Multi-Camera Object Re-Identification using Hierarchical Neural Networks

• DOI: 10.1109/islped52811.2021.9502480
• 发表时间: 2021-06
• 期刊: 2021 IEEE/ACM International Symposium on Low Power Electronics and Design (ISLPED)
• 作者: Abhinav Goel;Caleb Tung;Xiao Hu;Haobo Wang;James C. Davis;G. Thiruvathukal;Yung-Hsiang Lu