Better protection for passengers and staff
You receive early indications of visible violence, weapons, falls or motionless people.
With AI video analytics for public transport, you automatically detect defined visible events, evaluate them based on your rules and specifically inform your control centres, transport operations or security teams.
Initial position and goal
In your buses, trams, subways and regional trains, changing passenger numbers meet limited and constantly moving spaces. Drivers cannot simultaneously monitor traffic, passenger change and all areas of the vehicle.
In your buses, trains and stops, passenger numbers and situations change within a short time. AI video analytics detects visible events, evaluates them according to your rules and supports your control centres, transport operations and security teams in a quick, traceable response.
You receive early indications of visible violence, weapons, falls or motionless people.
You can automatically report visible acts of damage in vehicles and stops for review.
They record passenger numbers, entry and exit points as well as high personal densities in defined areas.
You can narrow down relevant records based on vehicle, line, stop, camera, time period and visible features.
Solution approach – Testable scenarios
You determine which visible situation should be identified, when it is relevant and which action follows. The AI detects visible events, but neither intentions nor emotions or the legal meaning of a situation.
They can show visible patterns of movement that indicate a possible physical discussion.
This gives your control centre or safety team an early reference to the affected vehicle and the corresponding camera area. This will help you to examine the situation and coordinate the further response.
The AI analyses exclusively visible movement sequences. It does not detect any intentions, emotions or causes of a conflict.
In suitable camera areas, you can have visibly worn or clearly recognizable weapons and weapon-like objects detected.
When detected, your responsible team will receive the camera view for immediate inspection. Further processing is carried out according to your defined safety and escalation process.
The detection depends on lighting, camera perspective, distance and visibility. It does not replace any assessment by trained security personnel.
They can be automatically informed if a person crashes visibly or remains motionless throughout a defined period of time.
Your drivers, control centre or safety team will receive an alert on the vehicle affected and the corresponding camera area. This allows you to check the situation early and coordinate the necessary help.
The analysis complements your emergency organisation. It does not replace first aid, emergency services or medical assessments.
You can monitor passenger compartments and stops for visible acts of damage.
These include, for example, repeated hitting or pedaling against doors, windows, seats, displays, ticket machines and other facilities, as well as visible smearing of surfaces.
If a corresponding motion pattern or a defined visible change is detected, your responsible team receives an alert with the corresponding camera view.
The AI detects only visible actions and changes. It does not assess the intention of any person, the legal meaning or actual extent of any damage.
You can record entry and exit at defined vehicle doors and evaluate the visible utilization of individual vehicles or vehicle areas.
The data supports you, among other things:
The camera-based count supports your operational planning. It does not replace prescribed capacity controls or vehicle safety systems.
In busy vehicles, at doors or in joint and entry areas, passengers can condense within a short time.
You can set threshold values for defined vehicle areas. If a high density of people is detected, your control centre or operating system can be informed automatically.
The AI evaluates only visible numbers of people and spatial distributions. It does not assess the mood or behaviour of individual passengers.
You can automatically recognise newly parked or unattended items over a specified period of time.
This will help you monitor passenger compartments, entry areas and stops. The event can be transmitted to your responsible team with the vehicle, camera, time and images.
The video analytics does not assess whether an object is dangerous. This evaluation remains the task of your security teams.
You can detect visible smoke or signs of fire in vehicles, stops, depots and other monitored areas in camera based.
When detected, your responsible team receives an alert with the associated camera view and can check the situation.
The video analytics complements your existing fire alarm and security systems. It does not replace any prescribed smoke detectors, fire detectors or operational fire safety measures.
You can monitor your door areas, mid-points, transitions and emergency exits for visible obstacles.
If these areas are blocked by baggage, bicycles, strollers or other items, the event may be transmitted to your responsible team.
The analysis complements your operational controls. It does not take over safety-critical control of vehicle doors and does not replace terminal protection, door-release or locking systems.
You can record the number of people, directions of movement and high densities at your stops, waiting areas and transfer points.
If defined thresholds are exceeded, your company will receive an alert and can plan additional vehicles, passenger information or safety forces if required.
The achievable accuracy depends on the camera perspective, visual conclusions and the spatial demarcation of the stop area.
You can see at which doors and stops particularly many passengers get on or off.
This gives you a better basis for your stop planning, vehicle planning and passenger steering.
Personal counting and direction of movement are evaluated anonymously, provided that no identification is required for the respective application.
You can provide driver’s cabins, technical areas and other non-public vehicle zones with individual rules.
If a person enters a defined area or access remains open for permitted situations, your control centre can be automatically informed.
The video analytics complements your existing structural and technical access security. It does not replace them.
You can monitor your vehicle halls, parking areas, workshop entrances, loading areas and fencing facilities outside the operating hours.
People or vehicles in defined restricted areas can be automatically detected and reported to your responsible team.
Zones, time slots and dwell times help you to distinguish regular operations from events requiring auditing.
Employees, buses, trains, service vehicles and machines meet in their company depots and workshops.
You can set up and detect virtual security zones when people enter closed driving areas or when vehicles are close to close their centre.
The video analytics complements your occupational safety and traffic rules. It does not replace instructions, route guards or prescribed controls.
In suitable workshop and operating areas, you can check whether visibly recognizable protective equipment is worn.
This may include, for example, safety vests, protective helmets or other clearly visible protective clothing.
The detection complements your occupational safety processes. It does not replace instructions or in-person checks.
With defined visual inspection tasks, you can determine after entering your depot whether visible items have been left behind or whether there have been clearly recognizable changes in the passenger compartment.
This allows you to respond to cleaning, object and maintenance processes in a more targeted manner.
The visual inspection only recognises clearly visible states. Covered items or invisible damages cannot be recorded.
After an event, you can narrow existing video recordings by vehicle, line, stop, period, camera, object type, direction of movement or visible personal features.
Your control centres and security teams do not have to manually search all records. Relevant sequences can be found faster and secured for further processing.
Video search supports you in the reconstruction of a process. The final evaluation remains the responsibility of your responsible teams.
From camera view to security process
You connect every visual detection with clear rules, designated teams and a defined operational workflow.
Existing or new cameras capture the areas defined for the application.
The AI evaluates the required visual features, objects, movements or actions.
zones, time frames, dwell times and other rules determine when a situation is classified as a relevant event.
When the system detects a defined deviation, it informs the responsible teams, control centres or connected systems.
Further processing takes place after a defined operational workflow, for example through visual verification, feedback, intervention or escalation.
The event can be documented in a traceable way with time, camera location, image, video clip and processing status.
Integration into your existing systems
You can transmit detected events via interfaces to your existing security, operating and management systems.
By connecting to the vehicle number, line, direction of travel, position and time, your control room will receive the necessary context for further processing.
Safety-critical decisions Your driving operations must not be carried out automatically based solely on the basis of a video analytics.
Your concrete added value
The benefit arises when you reliably detect defined visible events and connect them to the right safety or operation process.
You receive early indications of violence and visible weapons and check falls and emergency situations more quickly.
They detect visible vandalism and damage actions in vehicles and stops.
They better monitor passenger numbers and vehicle utilization rates, recognise high personal densities and record entry and exit movements.
You automatically recognise abandoned objects and transmit images to your responsible team.
They complement your fire alarm systems with an additional visual detection layer.
They monitor door areas and escape routes and protect driver’s cabins and operational areas.
Secure depots and operating yards outside service hours while supporting occupational safety processes.
You document relevant events in a structured manner and search existing video recordings more efficiently.
You can continue to use your existing vehicle and stop cameras and adapt rules to vehicles, lines, areas and operating hours.
Privacy by Design and Technical Limits
You set the video analytics according to the principle Privacy by Design around. You will only analyse those camera areas and visual features that are required for the specified purpose.
Toilets, staff rest areas and other areas of particular privacy will include you from the video analytics.
You coordinate the implementation with the applicable data protection regulations, permits, labour law requirements and your operational security requirements.
The relevant area, the action or the object must be sufficiently visible from your camera perspective.
Resolution, compression and frame rate must fit the visible object size and the expected movement.
Changing lighting conditions, reflections and backlight are checked with representative images.
Vehicle movements and vibrations can influence the detection. We check the image stability during the ride.
People and luggage can temporarily cover relevant movements or objects.
Size, seating arrangement, gears and joint areas determine the fields of view of the cameras.
High densities of people can make it difficult to identify and count individual passengers.
The quality of the vehicle and mobile connection determines how events are transmitted. Local processing or caching depends on the architecture you have chosen.
Detection and transmission time must fit your operational reaction process.
In the detection of violence and vandalism, the relevant movements or changes must be clearly visible in the camera view. Close passenger compartments, strong vehicle movements and mutual coverings can make it difficult to classify.
The AI detects exclusively visible features, objects and movements. It cannot judge any intentions, emotions, possible alcohol or drug influence, nor the actual danger or legal significance of a situation.
The video analytics does not replace vehicle-side safety, door, fire alarm or emergency call systems and no prescribed operational controls.
Each security alert must be checked by your responsible team and processed according to your defined safety and operation process.
Privacy by Design
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.
For private operators in Switzerland address: FDPIC guidance on video surveillance.
Frequently Asked Questions
In many cases yes. The decisive factors are image quality, perspective, coverage and technical connection. We test your existing cameras for the desired application.
Basically yes. However, vehicle movements, vibrations, changing lighting and connection interruptions must be taken into account in the technical design.
The AI can detect defined visible movement patterns that indicate a possible physical altercation. The final evaluation is carried out by your responsible team.
The video analytics can detect defined visible acts of damage or changes. The prerequisite is that the relevant action or change in the camera view is clearly visible.
You can count people at defined vehicle doors or within specified areas. The achievable accuracy depends on the camera position, crowd density and visual conclusions.
No. You do not need biometric identification for the applications described. Visible events, objects, movements and numbers of people are analysed.
It depends on the technical architecture chosen. Events can be processed locally or temporarized depending on the implementation and transferred after the connection has been restored.
No. It supports your teams with automated alerts and targeted prioritisation. Decisions, controls and interventions remain the responsibility of your responsible teams.
Your specific application
We test your cameras, vehicle types, operations and integration requirements and develop a suitable detection, integration and response logic.
