How Visual AI Makes the Modern Workplace Safer

demo

Whether on a factory floor, an offshore oil platform, or a gas-fired power generation plant, effective Health, Safety, and Environmental (HSE) management programs are critical not only to the health and well-being of employees but also to ensuring productivity and overall performance. However, with every new project, work site, and piece of equipment comes a wide variety of safety risks, leading to an ever-increasing demand for technology-enabled solutions. Recently, SparkCognition Vice President Cory Rhoads delivered a webinar in which he described the many opportunities artificial intelligence (AI) presents for company safety leaders charged with addressing this wide-ranging set of challenges. 

 

HSE challenges in the workplace today

Throughout the U.S., nearly 3M workplace incidents and injuries occur each year across a wide range of industries, with manufacturing leading the way in these unsafe acts and accidents. This volume of events results in companies spending over $250B annually on health and safety concerns alone. Effectively managing and minimizing workplace accidents is not just about dealing with the personal, financial, and regulatory consequences of such accidents after they have occurred. Far more important is knowing how to avoid them altogether, meaning that health and safety specialists need to leverage actionable insights from previously unseen hazardous conditions and events to detect imminent risks of all different types in the workplace. And taking steps to minimize these accidents means being proactively notified of near-misses, unsafe acts, and potentially dangerous environmental conditions. 

 

“The cost of injuries range from just under $50K to well over $1M in the case of fatalities,” Rhoads noted. “On average, employers spend approximately $120K per safety incident. To make matters even worse, OSHA investigations can drive an additional $150K per incident.” 

 

Many companies today address workplace safety in a reactionary way. After suffering an accident, realizing a significant health and safety issue, or maybe a compliance issue with personal protective equipment (PPE), companies will implement some type of reactive process change or a new form of control. Sometimes they’ll undertake additional training to create greater compliance and focus on PPE. But these are costly approaches, and they’re all looking in the rearview mirror rather than being proactive. 

 

But leading HSE practitioners, rather than focusing on after-the-fact actions, are adopting a proactive approach that recognizes the importance of reducing the number of unsafe acts, thereby reducing the number of significant health and safety incidents. Gaining awareness of hazardous actions and conditions can be challenging, as they frequently go unreported, leaving organizations blind to the risks their workers take each day. 

 

There are two primary reasons for this dearth of information on unsafe acts and conditions: people and processes. On the people side, contributors include alert fatigue, long hours, frequently changing shifts or simple complacency. On the process side, it can be difficult, time-consuming, or impossible to go back and assess video footage for potential risks or unsafe events. 

 

Rhoads observed that there are more than 700M surveillance cameras actively recording workplace activities around the world today. 

 

“But,” he said, “the vast majority of these are just passively recording. They’re overwriting the tape and are really only accessible through supervised management requests, and then typically only after an event has taken place.” 

 

5 visual AI use cases that create a safer workplace

Advanced artificial intelligence systems like SparkCognition’s Visual AI Advisor can use an organization’s existing camera infrastructure, whether fixed, drone-mounted, or mobile phones. Visual AI is a form of machine learning uniquely equipped to tackle the challenge of ensuring safety in the workplace. This analysis is accomplished not simply by identifying specific images but by contextualizing what it sees, whether that be a missing piece of PPE, a path someone is walking, or an action a worker is taking (or is about to take). The key is identifying potentially unsafe practices and conditions before they become accidents. 

 

Work practice awareness

The first use case in which Visual AI can detect impending safety problems is in monitoring worker movements and actions. This includes detecting improper bending and lifting techniques that can lead to falls or short- or long-term musculoskeletal injuries. Such injuries contribute not only to lost work time but also potential worker’s compensation costs. With Visual AI, workers can be proactively alerted to unsafe actions through nudges to wearable devices or cell phones. Such alerts can suggest that workers take a break or even inform them about the incorrect ways they’ve been moving or lifting, possibly even automatically signing them up for appropriate training. In a worst-case situation, the system can place a 911 call for a worker who’s injured. At one food manufacturer, SparkCognition’s Visual AI was able to reduce the number of incidents by four times based simply on the ability to capture these unsafe acts and near-misses.

 

PPE compliance

Another important use case is adherence to guidelines on wearing of PPE, e.g., helmets, face masks, vests, safety shoes, hairnets, gloves, etc. The presence or absence of all this equipment can be automatically detected, allowing managers to coach workers in real-time and generate reports about PPE usage over time. This capability improves organizations’ ability to ensure compliance with internal policies/procedures and OSHA guidelines while contributing, as well, to food safety and other product quality processes.  

 

Vehicle and human interaction

The third use case involves vehicle management, particularly in instances where vehicles (piloted or automated) operate in close proximity to workers. Such vehicles could include materials handling equipment like forklifts, robotic vehicles, or any other types that pose potential hazards to humans. Visual AI’s ability to monitor and alert on vehicle speed, location, proximity to humans, etc., is critical to ensuring that people and machines are able to work safely and effectively near each other. An important point Rhoads brought up was AI’s ability to monitor multiple safety elements in a single image or video feed, e.g., a vehicle operating near a blind corner close to a worker who’s not wearing the requisite PPE (meaning that any of the five use cases described here can be applied individually or in combination).  

 

Machine guarding

The fourth use case Rhoads described involves machine guarding. With the market’s current labor shortages comes a persistent knowledge gap, meaning companies today don’t necessarily have the same level of skill in their work environments that they have enjoyed in years past. Monitoring the existence and proper placement of systems like chain and blade guards is thus critical to maintaining safety. Further, Visual AI systems can determine if the person being viewed is authorized to be in that area based on the level of their training, job description, etc.  In addition to monitoring for proper placement of guards, the system also sees if a worker’s hand or foot has crossed a safety line or painted boundary. In response to such instances, alerts can be provided in the form of klaxons, flashing lights, or, in worst-case situations, automatically shutting the equipment down. And it can all be done in milliseconds. 

 

Privacy matters

The final use case Rhoads highlighted was in the area of data privacy. “Companies want to be respectful of worker privacy,” he said, “while at the same time being responsive and responsible. To achieve these possibly conflicting goals, AI systems can anonymize video images by obscuring faces, badges, vehicle license plates, etc. With this capability, workers can be active participants in ensuring the safety of their workplace without fearing penalization.” 

 

A new, proactive HSE solution leveraging Visual AI

Leveraging AI-enabled video analysis is a powerful tool for ensuring HSE performance in the workplace. SparkCognition’s Visual AI Advisor provides a wide range of capabilities without needing on-staff data science expertise or resource-intensive manual analysis of video feeds. Our system today is deployed on over 100,000 active cameras worldwide. Users gain the described benefits through simple drag-and-drop interfaces that allow flexible operations while ensuring continuous awareness of safety events or conditions that might lead to problems later. 

 

Rhoads closed his remarks by describing a paradigm shift in which industry leaders are moving away from reactionary reporting of accidents to one in which accidents and unsafe events are mitigated proactively through the use of real-time visual image analysis, often using infrastructure already in place for managing productivity and product quality, thus providing a win-win capability. 

 

Learn more about our AI-enabled HSE solutions

Watch the complete webinar here: 5 Ways to Transform Worker Safety with Artificial Intelligence 

Latest blogs

Visual-AI-School-Safety-unsplash
Uncategorized
Campbell LeFlore

Visual AI for School Safety: Time to Act

SparkCognition Visual AI Advisor delivers a highly versatile, intelligent, always-on solution leveraging the power of computer vision and artificial intelligence to transform school safety from

Read More

SparkCognition is committed to compliance with applicable privacy laws, including GDPR, and we provide related assurances in our contractual commitments. Click here to review our Cookie & Privacy Policy.