I-Corps: Human Behavioral Models for Camera Placement and Video Analytics

物理空间被划分为基础设施、建筑和设施，为人类服务于各种目的。例如，在公共设施的设计中，走廊或通道的创建是为了方便人们从一端到另一端的流动。类似地，在部署视频摄像机网络时，放置摄像机是为了捕捉人类活动并获得可识别的信息。然而，在上面的例子中，在设计摄像头放置的基础设施时，只考虑了非常有限的观察到的人类行为信息。诸如可持续性、成本、美观等因素往往优先考虑。这个I-Corps团队认为，在设计设施或设计摄像机网络中的摄像机位置时，人类行为理解与其他因素同等重要。该团队已经开发出一种方法，可以预测任何基础设施中预期的人类运动行为，进而利用这种方法来通知摄像机的位置，以形成视频摄像机网络。该解决方案优化了捕获可分析人类数据的能力，同时确保了当前部署中使用的传统智慧，包括监控固定资产、进出方式和开放空间的能力。提出的解决方案提供了一个具有成本效益的摄像机放置设计，形成一个视频摄像机网络，生成可操作的数据，从而最大化部署系统的价值和投资回报率（ROI）。提出的技术核心包括人体运动行为的方法和模型。它包括基于其已知几何结构的基础设施中的人体运动行为的计算模型，以及在已知几何结构中模拟人体运动的计算模型。其基本方法利用心理学和社会学领域的人类行为概念，并采用计算机科学优化技术来预测人类行为。该团队认为，所提出的方法可以满足或缓解当前行业对有效监测的需求，同时降低成本，以实现可操作的数据。在实现上述目标的过程中，该团队将识别现有的视频监控系统提供商、设计人员和集成商；将进行客户访谈，验证他们的需求，并确定可以在最小可行产品（MVP）中开发的功能特性。团队还将确定开发原型MVP所需的开发资源和设施，如果可能的话，按照敏捷开发方法将开发的方法集成到原型产品中。

Physical spaces are compartmentalized to form infrastructures, buildings, and facilities that serve various purposes for humans. For example, in the design of a public facility, corridors or passageways are created to facilitate the flow of humans from one end to another. Similarly, in deploying a video camera network, the cameras are placed to capture human activities and obtain identifiable information. However, a very limited amount of observed human behavioral information is taken into consideration while designing the infrastructure of the placement of cameras, respectively, in the above examples. Factors like sustainability, costs, aesthetics, etc. tend to take precedence. This I-Corps team believes that human behavior understanding holds equal importance as other factors in designing a facility or in designing the placement of cameras in a camera network. The team has developed an approach that predicts expected human motion behavior in any infrastructure, which in turn is leveraged to inform the placement of cameras to form a video camera network. This solution optimizes the ability to capture analyzable human data while ensuring traditional wisdom used in current deployments that include the ability to monitor fixed assets, entry/exit ways, and open spaces. the proposed solution delivers a cost-effective design for camera placements to form a video camera network that generates actionable data and hence maximizes the value and return-on-investment (ROI) of the deployed system.The proposed technology core comprises of methods and models for human motion behavior. It includes a computational model of human motion behavior within an infrastructure based on its known geometry, and a computational model to simulate human motion in a known geometry. The underlying methods leverage concepts of human behavior from the fields of psychology and sociology, and employs computer science optimization techniques to predict human behavior. This team believes that the proposed methods can fulfill or alleviate current industry needs related to effective surveillance at reduced costs to realize actionable data. In realizing the above, the team will identify existing video surveillance system providers, designers, and integrators; will perform customer interviews and validate their needs and identify functional features that could be developed in a minimal viable product (MVP). The team will also identify the development resources and facilities necessary to develop the prototype MVP and, if possible, work towards integrating developed methods into a prototype product following agile development methods.