Aegis Vision · Oil & Gas · Construction · India · Live Deployment

Enhanced Worker Safety
in Oil & Gas Construction
with Automated Video Analytics.

400K+
Images analysed

9
Cameras deployed

6
PPE categories

2.3M
Workers lost globally / yr (ILO)

Worker Safety — A Global Challenge

The scale of the problem
demands a different approach.

Automated monitoring deployed where the risk is highest. Built on existing infrastructure — no new hardware required.

Jump to the solution →

Construction and related fields like Oil & Gas present elevated risks by the nature of the work and environment. According to the International Labour Organization (ILO), in 2020 alone an estimated 2.3 million workers globally succumbed to work-related accidents or diseases.

Beyond the human cost, the ILO estimates that work-related injuries and illnesses cost the global economy an estimated 3.3 trillion USD annually — encompassing medical expenses, lost productivity, disability payments, and replacement training costs.

The question for industrial operators is no longer whether to act. It is whether the technology is ready to act reliably at scale. This deployment answers that question directly.

Live System Output

What the system sees
on every frame.

Automated PPE detection across construction workers — helmet, gloves and boots compliance identified simultaneously in real time

Ground-level camera view (HIKVISION 4MP at 8ft) — helmet, gloves and boots compliance assessed per individual worker simultaneously across the full frame. Red bounding boxes flag non-compliance items in real time.

Why Manual Monitoring Fails

Three limitations
that technology can solve.

01
Scalability

As construction projects grow in size and complexity, manual monitoring becomes increasingly difficult to maintain across numerous sites. A single safety officer cannot be everywhere simultaneously — and the cost of adding headcount scales linearly with the problem.

02
Limited Coverage

Safety officers can only observe specific areas at a time, leaving blind spots where violations occur undetected. The most critical non-compliance events happen between inspection rounds — not during them — and leave no record.

03
Human Error

Manual observation is susceptible to fatigue and inattentiveness, potentially missing critical safety issues. AI detection accuracy does not degrade over a 12-hour shift. Human attention does — and that gap is where incidents occur.

Before & After

From periodic inspections
to permanent monitoring.

Before — Traditional approach
Safety officers conduct scheduled rounds — violations occur between them, undetected
Coverage limited to one zone at a time; blind spots on every site, every shift
Observation quality degrades over long shifts due to fatigue
No structured data trail — violations go unrecorded unless caught in the moment
Expanding to new sites requires hiring and deploying additional safety officers
After — Aegis Vision deployed
Every camera monitors every frame continuously — no schedules, no blind spots
All 9 cameras monitored simultaneously from a single centralised dashboard
AI detection accuracy is consistent across a full 12-hour shift without degradation
Every non-compliance event logged, timestamped and attributed by site and camera
New sites added to the same platform — no additional officers required
What the System Monitors

Six safety categories.
One platform.
Zero gaps.

Aegis Vision monitors six categories of PPE and safety compliance simultaneously — every worker, every camera, every frame. Non-compliance events are logged to MySQL with full site and camera attribution, surfaced to the dashboard in near real-time.

The system was integrated with the client's existing HIKVISION CCTV infrastructure. No new hardware. No procurement cycle. Processing live within days of configuration.

Safety Helmet

Head protection compliance assessed per worker, per frame, continuously.

Safety Gloves

Hand protection identified on all active workers in the camera field.

Safety Boots

Foot protection monitored throughout all working hours on site.

Safety Harness

At-height harness compliance monitored for scaffold and elevated work.

Job Area Barricading

Work zone demarcation and restricted area boundaries verified continuously.

Scaffolding Safety

Structural scaffold integrity and access compliance monitored at height.

Detection in Action

Harness violations. Zone breaches.
Detected in real time.

Safety harness non-compliance detected on scaffolding in real time

At-height detection — safety harness non-compliance flagged immediately on active scaffolding. System identifies the scaffold structure and worker harness status within the same frame.

Zone violation alerts triggered across construction sites

Zone violation detection — personnel entering restricted areas triggers immediate alerts. Deployed across both multi-level building-frame and ground-level excavation environments.

Panoramic camera coverage showing multiple workers tracked simultaneously across full site footprint

Panoramic camera view (HIKVISION 6MP at 20ft) — the wide-angle camera tracks multiple workers simultaneously across the full site footprint. Helmet, gloves and boots compliance assessed per individual in a single frame pass.

Technical Solution

Built on what
you already own.

No new cameras. No hardware procurement. No site rewiring. Integrated with the client's existing HIKVISION CCTV network and processing live footage within days of configuration.

Honest scope: This implementation uses cloud processing on AWS. Edge and hybrid options are available for sites with restricted connectivity or data sovereignty requirements — scoped individually per engagement.
YOLO AWS EC2 g4dn.xlarge NVIDIA T4 GPU FastAPI Celery RabbitMQ Amazon S3 MySQL
Camera Infrastructure
7× HIKVISION 4MP cameras at 8ft for PPE monitoring. 2× HIKVISION 6MP panoramic cameras at 20ft for zone violation monitoring. Client's existing hardware — zero new procurement.
AI Model
YOLO deep learning model trained on a labelled dataset of workers with and without required PPE — optimised for construction-site conditions, lighting variation and camera angles.
Cloud Processing Pipeline
Motion-triggered images pushed from IP cameras via FTP (vsftp) to AWS EC2 g4dn.xlarge. Custom cron job transfers to Amazon S3. Celery distributed task queue and RabbitMQ broker manage concurrent camera workload at scale.
Audit & Dashboard
FastAPI service exposes model inference. Every non-compliance event logged to MySQL with site name, camera ID, PPE category, timestamp and image reference — creating a structured audit trail no manual inspection could produce.
Results & Impact

Measurable outcomes.
No asterisks.

The deployment validated the platform's ability to operate continuously at scale across multiple concurrent camera feeds. Beyond volume, it produced something manual inspection never could: a structured, timestamped audit trail of every non-compliance event per site, per camera, per shift.

Primary outcome
400,000+ images processed for PPE non-compliance across all cameras and shifts — without a single additional safety officer.

The system ran continuously across 9 cameras, processing every motion-triggered frame through the YOLO model and logging all violations to the audit database in near real-time.

400K+
Images analysed
9
Cameras live
Safety culture
Improved worker awareness and accountability across all active sites.

Automated monitoring — and the knowledge that every camera sees everything — fostered stronger individual PPE adherence throughout the deployment period.

Continuous
Coverage
Data intelligence
Structured compliance data now informs targeted safety training interventions.

The audit trail identifies which zones, which shifts and which PPE categories carry the highest recurring risk — enabling proactive training decisions that manual observation could never support.

Real-time
Alert delivery

See it running.
On a real site.

Book a walkthrough →