Cemin is a copper and gold mining company with two operations in the Valparaiso region of Chile. More than 300 Cemin employees live and work in close proximity at each location, so Covid-19 posed potentially catastrophic risks. An outbreak among workers could spread fast with major impact on employee health, mining operations, jobs, and company revenue.
Protecting employees is an important part of the culture at Cemin. In the early days of the pandemic, the company acted quickly to develop a comprehensive system to safeguard employees, including mask wearing and social distancing requirements, health screening questionnaires, and education and training programs. While the company had strong protocols in place, there was no efficient way to monitor compliance, identify problem areas, and track anomalies. It was essential for safety that people followed the guidelines, especially in public spaces such as cafeterias, break rooms, transportation areas, and during work within each of the mines.
Lima-based Quipu Consulting, a technical solutions provider specializing in image processing, artificial intelligence and other next-generation technologies, led the project. Quipu recognized early on how Ella real-time video search and alerting used for tailgating and other security functions could easily help with Covid-19 mitiga- tion regulations and compliance. Because Ella is a Software-as-a-Service platform that uses existing cameras, it took less than 45 minutes to “light up” cameras at Cemin’s remote mining locations without being on premise. Cemin did not need to purchase any additional hardware and wait for delivery and installation. The Ella Box VM runs on their existing VMware ESXi servers.
Ella's real time video search immediately detects areas where Covid-19 mitigation protocols are not being met. Entering simple natural language queries like “too close without mask” into the Ella search box reveals potential problem areas.
Ella uses AI and machine learning to turn regular 2D cameras into 3D sensors capable of detecting the distance between people. It anonymously tracks people as they move on a virtual 3D floor-plane grid — analyzing movement without storing any personally identifiable information (PIil. Measuring distance is safer than taking temperatures since people without fevers can still be contagious. And even when people are 6-feet apart, Ella automatically detects whether they are wearing masks.
Real-time dashboards track the rates at which people wear masks and maintain social distance to measure policy effectiveness. Ella “Event Streaming” pushes only important events directly to the Cemin team, who view alerts on their desktops or phones. “Event stacking” condenses and summarizes contiguous events, so time isn’t wasted reviewing unimportant or acknowledged events.
Security best practices such as TLS encryption with Perfect Forward Secrecy and 256-bit AES encrypted storage meet the most stringent IT requirements. Remote access is secure with no open in-bound network ports, and Ella operates securely by isolating the camera network from the corporate network.
Ella’s usage-based pricing makes actively monitored safety affordable. The flexible usage model also enables resources to be shifted precisely where and when needed. Query filters like “human or vehicle” optimize ROI by monitoring only the events related to the risks being mitigated. Ella scales to any size, so there Is no limit to the number of cameras. Storage is equally scalable and can be retained on-premise or in the cloud for as long as required.
At the outset of the Covid outbreak, Cemin’s board of directors acted quickly, giving the onsite management team only one week to implement a system to monitor and report Covid-19 mitigation strategies. Once Ella was selected to provide AI-driv- en intelligence, Quipu identified cameras in the most critical areas. After several days of analysis and planning, Quipu deployed Ella without ever being on site. It took less than 45 minutes to activate all cameras and be completely operational.
Quipu provided critical guidance to improve the accuracy of tracking and classification. Initially, there were false positive "no mask" detections in dining areas because people took off their masks to eat. Quipu added exclusion zones covering the transparent plexiglass barriers in those seated dining areas to reduce those false positives. Ella also deployed AI model updates weekly based on examples donated by Quipu to improve accuracy in each scene.
Real-time alerts accessible from any browser indicate issues as they happen. Automatic dashboards provide comprehensive insight and enable communication between team members for immediate response. The Cemin team uses dashboard metrics to manage compliance and to report weekly to its corporate board. “Our board of directors asked for reports on social distancing and mask compliance rates the week before Chilean Independence Day,” said Jose Borroni, Cemin’s chief information technology officer. “Ella made it easy. The automated real-time data and reports even link to the corresponding video events for quick review.”
More than 364K events per week are automatically monitored with real-time dash- boards for immediate detection and visibility into problem areas. The team also uses video clips as an educational tool. “Sharing video with the team was a great way to remind people,” said Marcos Joo, corporate project manager at Cemin. “It’s so easy to forget that something previously so harmless — like sharing a bag of olives — creates risk. Ella’s automated vigilance made it easier for everyone to view mask and social distancing compliance as a shared responsibility.”
Halting operations for even one month across all mines would have cost the company more than $4 million. By acting quickly with technology that puts cameras to work to understand the problems that warrant human attention, Cemin was able to sustain operations with no business interruption and protect employee health and well-being. The system also gives peace of mind. As Borroni adds, “The Ella system instills confidence in our team that we are keeping people safe while continuing operations.”
The system worked so well that Cemin, with Quipu’s support, is expanding its use for applications in plants and mines. For example, Cemin plans to train Ella to detect mining equipment and PPE like headlamps, hardhats and gloves that are required safety equipment. Cemin is also exploring Ella to evaluate interactions between people and machinery. “Deploying Ella for Covid-19 mitigation has shown that the same technology can help us identify ways to improve operations and productivity as well as protect employees. ” The software-based platform makes it easy to respond to multiple challenges addressed by AI visual identification and analysis — quickly and economically.