95%
PKN Orlen, bitumen tanker loading
PPE detection on existing cameras at the loading bays. Compliance reached 95% after six months and total alerts fell 85%.
PPE detection
PPE detection on every camera, every shift
Surveily checks that helmets, hi-vis, gloves and face shields are worn where they are required, on the CCTV you already have. Rules are set per zone, and each alert shows where and when PPE was missing.
Schedule a demo
Only 30 minutes.
No commitment required.
Software only.Runs on ~95% of standard IP cameras.One GPU server per site.
Category: 'Rocketship: Innovations in GenAI'
Trusted by
In short
PPE detection software uses computer vision on existing CCTV to check whether required protective equipment is worn in a zone.Surveily checks for helmets, bump caps, gloves, hi-vis vests, face shields, hearing protection, face masks, safety glasses and protective sleeves, each set per camera and zone. It detects a condition, not a person. No facial recognition or biometric identification is used.
95%
PPE compliance after 6 months
PKN Orlen refinery, bitumen tanker loading
85%
fewer total alerts
Same site, same programme
73%
fewer PPE alerts
Global steel producer, Polish site, 6 months
The problem
A good audit score does not mean PPE is worn
People behave differently on the day of an audit.
A manual audit covers one area on one day, often while people know it is happening. The score can look high while the next injury report still says required PPE was not worn.
At PKN Orlen's bitumen tanker loading, with material above 200°C and a real burn risk, continuous checks took PPE compliance to 95% in six months, with 85% fewer alerts. The change came from daily data and coaching, not from penalties.
Detection modules
Nine PPE types, set per zone
Helmet
Hard hat worn where head protection is required.
Cap / bump cap
Bump cap worn where it is required instead of a hard hat.
Gloves
Required gloves worn, including anti-cut types.
High-visibility vest
Hi-vis clothing worn in traffic and shared zones.
Face shield
Face shield or visor worn against splash, spark or impact.
Hearing protection
Ear defenders or plugs worn in high-noise areas.
Face mask
Respiratory or hygiene mask worn where required.
Glasses
Safety glasses worn in eye-hazard zones.
Sleeves
Protective sleeves worn where arm protection is required.
Rules are set per zone, so a helmet rule does not fire in the canteen.
How it works
What happens after missing PPE is detected
01
Detect
Surveily checks PPE on every configured camera and zone, all the time.
02
Prioritise
Severity depends on the zone. Missing hearing protection in a high-noise area comes before a missing glove at the door.
03
Alert
The right supervisor gets a de-identified clip by email, SMS or Microsoft Teams.
04
Coach
Weekly trends by zone and PPE type. The conversation is about the place and the pattern, never a named person.
Ask us how PPE detection would work on your site.
Schedule a demo
Only 30 minutes. No commitment required.
Measured impact
Results at sites already using PPE detection
73%
Global steel producer, Polish site
PPE alerts fell 73% over six months.
Before, we felt like safety rules were just another box to tick. Now, we see how data helps us stay safe every day. The AI system makes it clear that safety isn't about penalties, it's about safety culture.
EHS Manager, PKN Orlen
Privacy and security
PPE alerts show what is missing, not who
A PPE event records that a helmet was missing in a zone at a given time. It does not record who. People are de-identified by default, before anything is stored or displayed, and raw footage is analysed in working memory.
- No facial recognition
- No biometric identification
- No worker scoring or ranking
- No automated action against any worker
- De-identified before storage or display
- Full-body blur available
De-identified
What our customers say
European sites in heavy industry, chemicals and building materials, all running Surveily on the cameras they already had. No new hardware went in.
“
Surveily AI solutions have significantly supported our safety efforts - particularly in preventing near-miss incidents involving forklifts, maintaining order on vehicle routes and pedestrian pathways, and eliminating the use of mobile phones in restricted areas. The system provides us with real-time insight into on-site activities and enables quick response to potential hazards. It’s a tangible improvement in both operational visibility and our safety culture, which we highly value.
Piotr Łuksza
EHS Manager, Saint-Gobain Innovative Materials
“
Employee safety is our top priority - especially in the demanding environment of chemical manufacturing. Implementing Surveily’s systems significantly enhanced our oversight in critical areas. Their cutting-edge technology enables faster threat detection and real-time data analysis, resulting in a measurable improvement in workplace safety. We value Surveily’s professionalism and are proud to partner with them in building a work environment that meets the highest industry standards.
Radosław Studziński
HSE Coordinator, Reckitt Benckiser
“
We gained significantly better visibility and control over our expansive industrial site. Their innovative monitoring technology allowed us to effectively track vehicle movement and reduce hazardous driver behavior - particularly regarding the use of personal protective equipment and adherence to speed limits. As a result, we’ve enhanced safety for everyone working on-site. We firmly believe that partnering with Surveily is an investment that delivers tangible safety benefits.
Patrycja Szot
HSE Manager, Trzuskawica S.A. | A CRH company
Questions about PPE detection
Can we set different PPE rules per area?
Yes. Each PPE type is set per camera and per zone, including anti-cut gloves. A rule only fires where that PPE is required.
Does it identify who was not wearing the helmet?
No. The event records the missing PPE, the zone and the time. People are de-identified by default, before anything is stored or displayed, and no identity is attached.
How accurate is it?
Accuracy is measured per camera and reported to you as part of the service. During the pilot we validate results openly against our calibration target.
How long until it is live?
A typical deployment takes around eight weeks. We assess your cameras, choose the PPE rules per camera and install the edge server.
Can PPE data be used for disciplinary action?
Surveily never takes automated action against a worker, and events carry no identity. A person reviews every response. You are the data controller and decide how trends are used, in line with your works council agreement where you have one.
Check PPE on the cameras you already have
In 30 minutes, an engineer shows you Surveily on real, de-identified footage and tells you what it would detect on your site.
Schedule a demoOnly 30 minutes. No commitment required.
You remain the Data Controller. Surveily is the Processor.
