Detecting Threats earlier
Recognise visible weapons, violence, falls and other security-related events.
Retail and stores
AI video analytics forSupermarkets
AI video analytics automatically detects defined security events and visible operating states. Your team receives targeted advice, can react more quickly and does not have to observe all camera images permanently.
Detecting Threats earlierBetter control of customer flowsControl store conditions
Initial position and goal
Intelligently evaluate security, customer flows and visible store conditions
High customer frequency, movements of goods, cash register and logistics meet in your supermarket. At the same time, sales areas, entrances, self-service checkouts, storage areas and parking areas must be reliably monitored.
High customer frequency, movements of goods, cash register and logistics meet in your supermarket. AI video analytics detects visible security events and operating states, informs your team and provides data for faster responses and more efficient processes.
Better control of customer flows
Capture visitor numbers, queues, stay times and the utilization of defined areas.
Control store conditions
Check shelf gaps, blocked paths, open doors and other visible states regularly.
Incidents faster to investigate
Browse video recordings specifically for people, objects, vehicles and directions of movement.
The most important areas of application – Other useful applications
Use AI video analytics in your supermarket.
You determine which visible situation the system should detect, when it is relevant and which action follows. The AI detects visible events, but neither the intentions nor the legal meaning of a situation.
Weapon Detection
Detect visible weapons and threat situations
The video analytics can detect clearly visible firearms, knives and comparable weapons in defined camera areas.
If a corresponding object is detected, it is possible to automatically:
- Your staff will be informed discreetly
- Your internal security unit will be alerted
- the relevant camera view is displayed
- a short video clip
- a connected control centre will be notified
- Start a defined alarm process
The analysis provides an early note. It neither recognises the intention of a person nor does it decide on any further action independently. The event should be checked by your staff or control centre.
Peopleal protection
Recognising violence and physical disintegration
Certain clearly visible movement sequences can be identified as a possible situation of violence or conflict.
This is relevant, for example:
- in the cash register area
- at self-service cash desks
- in the input and output range
- on sales areas
- on service counters
- in parking lots
- in staffing branches
In the event of a detected event, your client gets the camera location and an appropriate visual context. Verbal threats, covert actions and not clearly visible conflicts cannot be reliably detected in the camera-based.
Emergency
Recognising falls and motionless people
If the video analytics detects a fallen person or motionless for a specified period, your staff may be notified automatically.
The application is suitable for:
- Sales spaces
- Entrance areas
- Cash reserves
- Parking spaces
- Stairs and roller chairs
- Storage and personnel areas
- rarely entered parts of buildings
Your team can immediately check the situation and initiate a first aid or emergency process if necessary.
The analysis supplements your organisational safety measures. It does not replace personal assistance, first aid or emergency processes.
Slide ride
Detect visible liquids and slipping hazards
Larger and clearly visible fluid accumulations in the camera view can be automatically reported as an inspection event.
This will support you, for example:
- drinks departed
- broken packaging
- visible impurities
- wet entrance areas
- Liquids against refrigerated cabinets
- Contamination in traffic routes
Your staff will receive the affected area and can arrange for cleaning or protection.
Small or colourless liquid spills, or spills obscured by people and trolleys, are not reliably visible in every condition. The analytics therefore does not replace regular inspection rounds.
Cash management
Detect queues at cash offices
The video analytics can capture the number of individuals and the length of defined queues.
You can set rules by which your team is informed, for example, if:
- a certain number of people is exceeded
- a queue reaches a traffic route
- the estimated waiting time increases
- Only a few cash registers are open
- individual cash registers are significantly more strongly utilised
This allows you to open additional cash registers directly and use your staff more tailored to your needs.
The calculated waiting time is an estimate based on visible movement and depends on the camera angle, checkout type and customer behaviour.
Self-service funds
Targeted support for self-service funds
In the area of your self-checkout area, video analytics can capture occupancy, queues and defined movements.
In conjunction with your POS system, events can be triggered if:
- a checkout is blocked for a longer period of time
- a technical or manual review alert
- a person leaves the area after an open operation
- Several people are staying at a single cash register
- a defined output without a completed operation is done
The video analytics alone does not reliably detect whether each item has been correctly recorded or paid for. Only the connection with cash register and transaction data enables a targeted examination. The final evaluation is done by your staff.
Visitor counts
Recording visitor numbers and utilization
Incoming and outgoing people can be counted anonymously. This will provide you with information about:
- Visitor numbers
- current store spread
- heavily frequented times of day
- Capacity for individual sales areas
- Seasonal changes
- Effects of actions
- Ratio of visitor numbers and cash utilisation
The evaluation can be carried out without notification by writing. It supports your personnel planning, area control and operational evaluation.
Customer flows
Analyse customer flows and dwell areas
Movement directions and lengths of stay can be evaluated anonymously. This allows you to see which areas of your branch are strongly or weakly frequented.
Possible evaluations are:
- frequently used walking paths
- Bottlings on the sales floor
- Duration of stay in defined zones
- Use of action areas
- Distribution of customers via the branch
- Movements between entrance, sales area and cash register
From this you can derive information for your route guidance, surface design and product placement. The data describe visible movements and not the interests, intentions or personal characteristics of your customers.
Shelf availability
Visual checking shelf gaps and goods availability
In suitable camera perspectives, it can be regularly checked whether defined shelves are visibly filled, partially empty or completely empty.
For example, your team can inform the analysis if:
- a fixed shelf zone empty
- an action area should be refilled
- a product display visibly mined
- a planned presentation of goods is missing
- larger gaps over a defined period of time
Reliable product-level detection depends on shelf layout, packaging, lighting and image resolution. Visual inspection complements your inventory management system but does not replace inventory control.
Protected areas
Securing cash, storage and personnel areas
Virtual zones can be placed over protected areas and linked to individual rules.
For example, the analysis can detect whether:
- enters a person’s staff area
- a warehouse access from defined times
- people staying in a health insurance office
- a delivery area is entered without authorisation
- one door stays open longer than provided
- after closing business, motion in the store is detected
Time slots, length of stay, direction of movement and other rules can be adapted to your operational processes.
Event check
Check theft suspected cases more quickly
After a reported incident, relevant video material can be limited by time, camera, personal features and direction of movement.
For example, you can:
- link a transaction time with video material
- investigate a reported loss of goods
- Moves between shelf, checkout and exit
- Determine an affected branch zone faster
- Backing up relevant video clippings
The AI does not determine whether the theft has been committed solely by means of the camera view. It supports your staff in research and visual exams.
Property security
Monitor defined goods and objects
Sufficient and clearly visible objects can be monitored for distance or change.
This is suitable for:
- high quality product displays
- sealed container
- Presentation stand
- Transport boxes
- technical devices
- Cleaning machines
- Vending machines
- Outdoor equipment
The video analytics cannot assess whether a withdrawal is entitled. A message merely indicates the visible change to your staff.
Door and refrigerated cabinet control
Detect opened doors and refrigerated cabinets
In suitable camera perspectives, it can be checked whether doors, gates or sufficiently visible refrigerated cabinets are open over a defined period of time.
For example, you can monitor:
- Storage doors
- Supplier access
- Frozen chest
- Cold room doors
- Doors to personnel areas
- Emergency exits
- Rolling gates
The analysis detects the visible opening state. It does not measure temperature and does not replace prescribed temperature control or technical door sensors.
Branch states
Control visible cleanliness and branch conditions
Defined areas can be regularly checked for visible states.
These may include:
- Waste in transportation
- Overcrowded collection points
- parked boxes
- blocked cleaning areas
- disordered action areas
- empty shopping carts
- visible contamination
- Unremoved means of transport
The specific reliability depends on how clearly the desired state in the camera view is recognizable.
Carriages and baskets
Count shopping cart and shopping baskets
Shopping trolleys and larger shopping baskets can be counted in defined areas.
This allows you to detect, for example:
- empty shopping carts
- wagons parked incorrectly
- blocked return areas
- High demand in the entrance area
- Distribution between different collection points
This will enable your team to respond more specifically and improve availability for your customers.
Visual Fire Detection
Detecting smoke and fire early
Visible smoke or flames can be detected in the sales room, warehouse, delivery area, car park or waste collection points.
When an event is detected, the system can automatically inform your responsible team and provide the appropriate camera view.
The video analytics supplements your fire alarm and security systems with a visual detection layer. It does not replace any prescribed fire detectors or fire safety measures.
Abandoned Object Detection
Detect items left behind
The system can automatically report newly placed objects or unattended items over a defined period of time.
Your team will receive the camera location, the time and a suitable image or video clip. This allows the situation to be checked quickly without having to observe all camera images permanently.
Flight and traffic routes
Keep escape and traffic routes free
Defined escape, rescue and traffic routes can be monitored for visible obstacles.
These include, for example:
- Emergency exits
- Escape routes
- Main courses
- Access to fire extinguishers
- Rescue rides
- Delivery routes
- Areas before technical facilities
If the system detects pallets, roll containers, shopping carts, deliveries or other obstacles, your staff can be automatically informed.
Goods acceptance
Monitor delivery zones and goods receipt
Delivery vehicles, employees, pallets and means of transport meet in your goods receipt.
The video analytics can be detected, among other things, if:
- People enter a locked work zone
- Keep vehicles out of size planned areas
- Roll container blocking traffic routes
- Doors or gates remain open
- Deliveries to be left in a defined zone
- critically approaching people and vehicles
The rules are adapted to your delivery processes, security specifications and camera perspectives.
Vandalism Detection
Recognising vandalism
Visible condition changes or defined movement sequences on facades, machines, shopping trolley stations, parking facilities or technical facilities can be reported as review events.
The CI cannot clearly determine vandalism in every situation. However, it can document relevant visible processes and draw your staff’s attention to it.
Parking spaces
Monitoring parking areas and access points
video analytics can count vehicles, evaluate parking areas and monitor defined restricting zones on your customer parking lots.
Possible applications include:
- Registration of parking space occupancy
- Detection of blocked access points
- Monitoring of delivery and fire service departures
- Detection of vehicles in restricted areas
- Evaluation of the parking period
- Documentation of entrances and exits
Automated parking management requires clearly defined rules and, if necessary, connection with other systems.
Number plate recognition
Entering trademarks legally permissible
Vehicle licence plates can be read automatically at private entrances and exits.
This will support you, for example:
- the access control
- Parking management
- the search for an incident
- the occupancy of delivery vehicles
- the management of eligible vehicles
- the comparison with released reference lists
Trade number plate recognition and list matching are only used where there is a suitable legal basis. Automatic hits must be checked by authorised personnel before further action.
Video search
Incidents faster to investigate
After a security event, you can search extensive video archives based on defined criteria.
Possible search criteria are:
- Period
- camera or branch area
- People
- visible clothing features
- objects carried
- Shopping cart
- Movement direction
- Vehicle type
- Vehicle colour
- Licence Plates
This makes it possible to limit relevant video material more quickly without having to view all recordings manually.
How it works
From camera view to security process
Record
Your existing or new cameras capture the areas defined for the respective application.
Analyse
The AI evaluates the required visual features, objects, movements or changes in their state.
Evaluating
Zones, time slots, dwell times, number of people and other rules determine when a situation is classified as a relevant event.
Information
In the event of a detected event, your employees, your control centre or connected systems are specifically informed.
Responding
Your team checks the situation and processes it according to a defined operational workflow.
Document
Your team can document the event in a comprehensible manner with time, camera location, image, video clip and editing status.
Operational integration
Integration into your existing systems
Detected events and evaluations can be transmitted to your existing security, cash register and operating systems via interfaces. This creates a consistent process from visual recognition to human verification to operational reaction and documented processing. Possible integrations are:
Operational added value
Your concrete added value
Detecting Threats earlier
Previous detection of visible weapons and dangerous situations
Check emergencies faster
faster reaction during falls and visible sliding rides
Control cash registers
More needs-based control of your cash registers
Reacting on queues
Shorter response times for growing queues
Understanding customer flows
better evaluation of customer flows and store utilization
Control store conditions
regular control of visible shelf and branch conditions
Securing operating areas
improved safeguarding of storage, cash register and personnel areas
Keep ways clear
faster detection of blocked escape and traffic routes
Relieving employees
Relief for your employees and security bodies
Accelerate research
faster search for reported incidents
Documenting events
Structured documentation of relevant events
Continue using cameras
Use your existing camera infrastructure
Centrally manage branches
Central management of multiple branches
Comparing sites
cross-site evaluations and comparisons
Data protection and compliance
Limit the video analytics to the required purpose.
The implementation is based on the principle Privacy by Design. The analysis only takes into account those image areas and visual features that your specified purpose requires.
When monitoring employees, linking them with transaction data, licence plate recognition, biometric procedures and list matching, a clear purpose limitation, a suitable legal basis and individually defined access and extinguishing concepts are required.
Technical limits
Image quality and conditions on site
The reliability of video analytics depends on the quality and position of your cameras as well as the conditions in your store.
Impairments may arise, among other things, by:
- concealed visibility areas
- People close together
- high shelves
- Shopping Carts and Goods Display
- Reflections in refrigerated cabinets
- changing lighting
- backlight in the entrance area
- very small products
- similar packaging
- short visible actions
- unfavorable camera perspectives
- Insufficient image resolution
Visible features instead of intentions
The AI detects visible features, objects and changes in the state. It does not establish any intentions and does not decide whether an act is entitled, suspicious or punishable.
Scans and transaction data
Reliable control of individual scanning processes or smallest products is not possible using classic overview cameras alone. Applications to self-service funds therefore usually require a connection to the cash register or transaction system.
Shelf gaps and visible states
Shelf gaps and branch conditions can only be evaluated reliably if the corresponding surfaces are sufficiently visible. Temperature, product quality, actual stock and technical system states cannot be reliably determined from the camera view.
human verification and security systems
Automated alerts should be checked by authorised personnel. The video analytics does not replace robbery indicators, fire detectors, intrusion alarm systems, temperature monitoring or prescribed organisational controls.
Privacy by Design
Check which data is processed for which purpose.
Depending on the application, the following measures are possible:
- Processing directly in your branch
- Data storage in Switzerland or within the European Union
- Event-based storage
- Limitation of storage times
- Role-based access rights
- Logging of access and editing steps
- Picularation of faces or people
- anonymous person counting
- Aggregation of customer flow data
- Hiding public roads and neighbouring areas
- split permissions for video, licence plate and event data
- regular deletion of no longer required data
- Facial recognition only in legally permissible and separately audited cases
- Human Verification of automated hits
For locations in the EU form the GDPR, especially Art. 5, 6, 25 and 35 and the EDSA video surveillance guidelines the basis of the privacy audit. In addition, consider the law of the respective Member State, in particular employee data protection and co-determination. The duties from the EU AI Act check for the concrete use of AI.
Frequently Asked Questions
Questions about AI video analytics for supermarkets.
Can we continue to use our existing cameras?
In many cases yes. The decisive factors, viewing angles, lighting and coverage of the desired area are decisive. For shelf analyses, licence plate recognition or special cash register zones, additionally adapted camera perspectives may be required.
Does video analytics automatically detect shoplifting?
Not reliably based on the camera view. The AI can detect defined visible events and speed up search within a reported incident. Your personnel must check whether the theft is actually present.
Can self-service cash registers be monitored?
The video analytics can capture occupancy, queues and defined movements. A connection to your POS system is usually required to check open or conspicuous processes.
Does the system not detect scanned goods?
A reliable determination is not possible solely based on an overview camera. Cash and transaction data, suitable camera perspectives and clearly defined testing rules must be combined.
Can shelf gaps be detected?
Yes, if the shelf area in question is sufficiently visible. The reliability depends on the packaging, shelf structure, lighting and camera perspective.
Can customers be counted anonymously?
Yes. Visitor numbers and utilization can be evaluated without named identification. For this purpose, the visual features required for counting are processed exclusively.
Can the checkout waiting times be measured?
The video analytics can capture the length of a queue and its movement. This can be used to deduce an approximate waiting time. The accuracy depends on the checkout process and the situation.
Does the AI detect discontinued liquids?
Greater and significantly visible fluid accumulations can be detected. Small, transparent or concealed liquids may remain undetected. Regular inspections therefore remain necessary.
Is facial recognition possible?
A facial recognition is technically possible. However, it is only used for clearly defined use cases in which it is legally permissible and proportionate. Each hit must be checked by authorised personnel.
Can people be searched without facial recognition?
Yes. A research can be done using visible features such as clothing colour, carried objects, direction of movement and stay time.
Can several branches be evaluated centrally?
Yes. Events, visitor numbers, queues, cameras and defined operations of multiple locations can be centrally combined depending on your infrastructure and authorisation concept.
Does video analytics replace your employees or security systems?
No. It supports your team in identifying, checking and documenting visible events. Required fire detectors, robbery detectors, intrusion alarm systems and operational controls remain required.
Your specific application
You want to identify visible security risks earlier, control your cash registers better or automatically check store states?
We check which applications can be reliably implemented with your existing camera infrastructure and develop an appropriate analysis, integration and operational architecture for your supermarket or your entire branch network.
