Computer Vision and IoT for Personalised Site Monitoring Analytics in Real-Time (CV-SMART) towards Behaviour-Based Safety

“一人死亡太多”（Donaghy, 2009）英国零伤害议程的口号表明，建筑工地不允许发生致命事故。现代卫生与安全问题只能从文化、社会和技术的角度来解决。我们从文化和社会的角度做了很多工作，但技术的角度却被大量忽视（尤其是数字技术的使用）。有必要仔细研究社会和技术系统是如何重叠的，以及日益增长的重叠如何影响建筑工地的卫生与安全。然而，目前的方法单独解决了卫生与安全管理的社会技术重叠问题，尽管有证据表明，如果没有有效的数字化以及适当的社会干预，就无法实现零伤害目标和卫生与安全最佳实践（国家统计局，2018年）。因此，该联盟将利用最先进的计算机视觉和深度学习（卷积神经网络和循环神经网络）来开发用于个性化现场实时监控分析（CV-SMART）的计算机视觉和物联网，以实现密闭建筑工地基于行为的安全。CV-SMART将自动检测、识别和实时跟踪各种相互作用的重型机器、建筑部件、现场活动和现场工人，以（i)执行H&S最佳实践；（ii）识别H&S风险和不安全的现场实践，如未佩戴安全装置（头盔、安全带和个人防护设备PPE）、拥挤的工作区域、重型设备的不当移动和不适当的工作位置）。CV-SMART将提供一个数字可视化平台，该平台将采用最先进的可视化技术，用于直观的用户交互和报告。CV-SMART将提醒现场工作人员、安全管理人员和现场管理人员有关H&S的最佳实践和即将发生的风险，并通过使用支持物联网的设备将人为干预降至最低。

The "One Death is too Many" (Donaghy, 2009) catchphrase for the UK zero-harm agenda shows that no fatal accident is admissible on construction sites. Modern H&S problems can only be solved from a combination of cultural, social and technical perspectives. A lot of work has been done from cultural and social perspectives, but the technical perspective has been massively ignored (particularly the use of digital technologies). There is a need to take a closer look at how social and technical systems overlap, and how the growing overlap influences H&S on construction sites. However, current approaches address the socio-technical overlap of H&S management separately despite the evidence that suggests that the Zero Harm target and H&S best practices cannot be achieved without effective digitisation in addition to appropriate social interventions (NBS, 2018). The consortium will, therefore, leverage state-of-the-art in computer vision and deep learning (convolutional neural networks and recurrent neural networks) to develop a Computer Vision and IoT for Personalised Site Monitoring Analytics in Real-Time (CV-SMART) for behaviour-based safety in confined construction sites.CV-SMART will automatically detect, recognise and track diverse interacting heavy machines, building components, site activities, and site workers in real-time to (i) enforce H&S best practices and (ii) identify H&S risks and unsafe site practices such as failure to wear safety gadgets (helmets, harness, and personal protective equipment PPE), congested work areas, improper movement of heavy equipment, and improper working positions). CV-SMART will provide a digital visualisation platform, which will employ state-of-the-art in advanced visualisation for Intuitive user interaction and reporting. CV-SMART will alert on-site workers, safety managers, and site managers of H&S best practices and impending risks at the forefront of onsite operations with a minimal human intervention using IoT-enabled devices.