SCC: Building Safe and Secure Communities through Real-Time Edge Video Analytics

SCC：通过实时边缘视频分析构建安全可靠的社区

• 批准号: 1831795
• 负责人: Hamed Tabkhi
• 金额: $ 189.75万
• 依托单位: University of North Carolina at Charlotte
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2018
• 资助国家: 美国
• 起止时间: 2018-10-01 至 2025-03-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1831795&HistoricalAwards=false
• 关键词: SCC; Building; Safe; Secure; Communities

The emergence of intelligent technologies is enabling a new era of connection between community residents and the surrounding environments, both in the United States and around the world. With the new wave of growth in urban areas, ensuring public safety is an essential precursor toward "smart" cities and communities. This project proposes a novel "intelligent" policing technology as a transformative solution to efficiently enhance law enforcement, while minimizing unnecessary interactions and maintaining resident privacy. The proposed technology offers a network of smart cameras that do not require continuous monitoring, but instead are trained to generate alerts on the spot in real-time. Since the cameras identify behaviors and not identities, they can reduce biases, minimize false alarms, and protect personal privacy. The intelligent policing technology will be co-designed and co-created with the direct help of community residents, neighborhood leaders, and local business owners, as well as agencies including the City of Charlotte, and local law enforcement agencies in Charlotte-Mecklenburg and Gaston counties. The proposed research makes fundamental advances in multiple areas from computer vision, computer architecture, and real-time edge computing, as well as criminology and community-technology interaction. It paves the path for bringing the recent advances in deep learning and data analytics to enhance the safety and security of communities without jeopardizing the privacy of residents. To this end, this project formulates social-technical advances to efficiently analyze and assist communities and governing agencies in making real-time, smart reactions. The project enables real-time vision processing near the cameras (edge nodes) and cooperative processing over the edge network. At the same time, the proposed research interprets, formalizes, and models public safety and security events to be machine detectable, reducing biases, and enabling broad-based community support and trust. By demonstrating the use of powerful emerging edge computing technologies, the project will highlight the applicability and adaptability of such technologies to tackle many community challenges and broader smart cities and cyber-physical systems (CPS) applications, including smart transportation and pedestrian safety. Additionally, the proposed community-based pilots will serve as exemplars to other communities across the nation.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.

智能技术的出现正在开启美国和世界各地社区居民与周围环境之间连接的新时代。随着城市地区的新一轮增长，确保公共安全是迈向“智慧”城市和社区的重要前提。该项目提出了一种新颖的“智能”警务技术作为变革性解决方案，以有效加强执法，同时最大限度地减少不必要的互动并维护居民隐私。拟议的技术提供了一个智能摄像头网络，不需要持续监控，而是经过训练可以实时生成警报。由于摄像头识别的是行为而不是身份，因此它们可以减少偏见，最大限度地减少误报并保护个人隐私。智能警务技术将在社区居民、社区领导和当地企业主以及包括夏洛特市、夏洛特-梅克伦堡和加斯顿当地执法机构在内的机构的直接帮助下共同设计和共同创建县。拟议的研究在计算机视觉、计算机体系结构、实时边缘计算以及犯罪学和社区技术交互等多个领域取得了根本性进展。它为利用深度学习和数据分析的最新进展铺平了道路，以在不损害居民隐私的情况下增强社区的安全和保障。为此，该项目制定了社会技术进步，以有效分析和协助社区和管理机构做出实时、智能的反应。该项目支持摄像机（边缘节点）附近的实时视觉处理和边缘网络上的协作处理。与此同时，拟议的研究对公共安全和安保事件进行解释、形式化和建模，使其可被机器检测，减少偏见，并实现广泛的社区支持和信任。通过展示强大的新兴边缘计算技术的使用，该项目将强调此类技术的适用性和适应性，以应对许多社区挑战以及更广泛的智能城市和网络物理系统（CPS）应用，包括智能交通和行人安全。此外，拟议的基于社区的试点项目将成为全国其他社区的典范。该奖项反映了 NSF 的法定使命，并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，被认为值得支持。

项目成果

EfficientHRNet: Efficient and scalable high-resolution networks for real-time multi-person 2D human pose estimation

EfficientHRNet：用于实时多人 2D 人体姿态估计的高效且可扩展的高分辨率网络

• DOI: 10.1007/s11554-021-01132-9
• 发表时间: 2021-08
• 期刊: Journal of Real-Time Image Processing
• 影响因子: 3
• 作者: Neff, Christopher;Sheth, Aneri;Furgurson, Steven;Middleton, John;Tabkhi, Hamed
• 通讯作者: Tabkhi, Hamed

REVAMP 2 T: Real-Time Edge Video Analytics for Multicamera Privacy-Aware Pedestrian Tracking

REVAMP 2 T：用于多摄像头隐私感知行人跟踪的实时边缘视频分析

• DOI: 10.1109/jiot.2019.2954804
• 发表时间: 2020-04
• 期刊: IEEE Internet of Things Journal
• 影响因子: 10.6
• 作者: Neff, Christopher;Mendieta, Matias;Mohan, Shrey;Baharani, Mohammadreza;Rogers, Samuel;Tabkhi, Hamed
• 通讯作者: Tabkhi, Hamed

Mez: An Adaptive Messaging System for Latency-Sensitive Multi-Camera Machine Vision at the IoT Edge

• DOI: 10.1109/access.2021.3055775
• 发表时间: 2024-09-13
• 期刊: IEEE Access
• 影响因子: 3.9
• 作者: Anjus George;A. Ravindran;Mat'ias Mendieta;Hamed Tabkhi
• 通讯作者: Hamed Tabkhi