GenAI-powered solution cuts EHS incidents by 80% for a leading Danish multinational organization
Share this case study on
About the client
With deep engineering roots, our client, a Denmark-based multinational organization, helps enterprises increase machine productivity, reduce emissions, lower energy consumption, and enable electrification. The client has operational hubs across the globe. The company powers customers to operate efficiently and lead the way toward a cleaner and sustainable future.
Business challenges
Even a single lapse at any operational hub could result in product contamination, a recordable workplace incident, or a costly operational shutdown therefore the client’s safety team was spending excessive time managing continuous and overlapping safety risks across shifts and plants.
The client’s existing EHS operations were not effectively preventing safety and compliance breaches and there were breaches that led to penalties or incidents. The safety approach relied heavily on manual monitoring and periodic audits, which often resulted in gaps in occupational hygiene and safety. To strengthen its EHS foundation, the client partnered with Aspire Systems to address the following challenges:
- Frequent breach of critical “no-go” zones, with pedestrians and machinery often crossing into each other’s paths.
- Unauthorized access to high-risk areas such as reactor zones and high-voltage sections posing an immediate threat of serious injuries and fatalities.
- Undetected liquid leaks – not attended for extended periods, prompting to significant occupational safety risks.
- Ensuring sufficient safety supervision during night shifts and maintenance shutdowns
Solution approach
Aspire Systems was selected by the client to design and deploy an AI-driven EHS solution that transformed reactive, manual oversight into proactive, automated protection. Together, we engineered an integrated computer vision platform spanning four critical safety domains.
We started with a detailed evaluation of the client’s EHS landscape, collaborating with safety and operations teams to assess CCTV coverage, plant workflows, and points of failure in manual monitoring. Rather than proposing a large-scale hardware upgrade, the solution leveraged the existing camera infrastructure, enhanced with edge AI gateways to enable real-time, on-site processing.
The scope of this project spanned the following primary focus areas:
- PPE and hygiene compliance — ensuring every worker is correctly equipped before contamination or injury can occur.
- Vehicle and pedestrian safety — eliminating man-machine proximity risks in shared operating zones.
- Restricted area and machine guarding — preventing unauthorized access to high-hazard zones and machinery.
- Housekeeping and 5S compliance — detecting leaks, obstructions, and unsafe behaviors that erode plant safety<.li>
By adopting a modular, camera-agnostic architecture, we enabled incremental rollout of each capability, demonstrating value in one domain before scaling to the next, without disrupting the client’s business-as-usual operations
To improve PPE compliance, we designed and implemented an AI-powered detection system that analyses live CCTV feeds in real time to verify the use of essential protective gear such as hairnets and safety vests. By leveraging YOLOv7 for high-speed detection, OpenCV and RTSP for video-stream handling, and cloud-based inference on Azure, we replaced manual checks with continuous monitoring and audit-ready insights
To enhance vehicle and pedestrian safety, we developed a solution that calculates the real-time distance between people and machinery using live video feeds, triggering alerts whenever predefined safety thresholds are breached. To help supervisors quickly validate risks and reduce false alarms, we integrated a generative AI layer that delivers clear, contextual alerts along with a short video snippet.
For restricted-area access and machine guarding, we moved beyond basic motion sensors by deploying a Vision Language Model that could understand context. The solution could distinguish between an authorized technician performing maintenance and an unauthorized operator entering a danger zone. By combining instance segmentation for precise boundary detection with intent recognition, we integrated an edge-based Python inference engine with plant systems to trigger Category 0/1 emergency stops and halt machinery in sub-second response time, orchestrated through Azure IoT Edge.
For housekeeping and 5S compliance, we applied advanced computer vision to detect liquid leaks in chemical pathways, identify blocked safety equipment, and recognize unsafe worker behaviour before it escalates into a hazard. The system issues localized audio alerts to prevent incidents in real time. This approach enabled the client’s safety teams to shift from periodic inspections to continuous 5S compliance monitoring.
Unified under a single edge-to-cloud architecture with centralized policy management on Azure, these solutions transformed fragmented, manual EHS processes into an always-on safety intelligence platform. The modular design also allows the client to scale proven safety capabilities across facilities worldwide.
Value adds
- Enabled real-time, continuous safety monitoring across the client’s existing CCTV infrastructure—without disrupting current EHS processes.
- Reduced reliance on manual EHS monitoring, allowing safety teams to shift their focus toward proactive risk prevention and strategic priorities.
- Significantly reduced manual EHS monitoring effort, enabling safety teams to focus on higher-value prevention and strategic initiatives.
- Enabled audit-ready analytics with automated compliance reporting.
Business impact
- 80% reduction in high-severity man-machine proximity incidents.
- 63% reduction in high-risk zone violations.
- 30% reduction in downtime and recovery of 150+ production hours per year.
- Near Zero-latency intervention, preventing mechanical-contact injuries even before they occur.
- 40% reduction in aggressive driving behavior with real-time driver coaching.
