Custom detection models
We examine existing models and train additional object classes or states if necessary. Representative data and separate test cases form the basis for evaluation.
← ServicesDevelopment and
Development and
Customisation. Built for your needs.
With development and customisation, we complement the functions that your Visual AI solution lacks for specific use.
This can be a special recognition, an approval step, or a custom user interface. We use appropriate platform components and develop the necessary extensions.
Build visual workflows for specific tasks
Develop operation and workflows appropriately
Keep changes testable and maintainable
Your entry
Where is the standard function not sufficient?
Do you want to detect a special condition, check results according to your own rules, or provide employees with a simpler interface?
We first clarify what can be configured and which function needs to be developed. A defined prototype makes the critical assumption verifiable early.
Service components
What we develop for you.
You can commission individual extensions or a coherent specialist application. We define requirements and acceptance criteria together.
Visual Inspection Tasks
We design inspection routines using vision-language models, or VLMs. Questions, permitted answers and plausibility checks are defined so that the results can be processed in a structured form.
Backend and APIs
We develop the required services and data models. This includes access rules, events, and interfaces for exchange with your platform or your specialised applications.
User interfaces
We design views for the specific task: such as a verification screen, configuration, or evaluation. Employees test early whether information and action options are arranged in an understandable way.
Rules and workflows
We map approvals, escalations, notifications, and follow-up actions. Responsibilities and exceptional cases are explicitly taken into account so that the process remains understandable even with incomplete results.
Tests and Versioning
We verify software functions with automated tests and visual recognition functions using agreed sample data. Versions, changes and known limitations remain traceable.
How we work together
Develop in verifiable steps.
You see interim results early and provide expert feedback before further functions build on them.
01
Determine Scope
Together we prioritise user tasks and describe the first deliverable feature. This includes data access, technical limitations, and concrete acceptance criteria.
02
Test prototype
We implement the core function. Your employees test it with appropriate examples. The feedback determines which assumptions need to be adjusted.
03
Expand application
The confirmed function is supplemented by operation, access rights, error handling, and interfaces. New versions are checked against the agreed criteria.
04
Handover release
We document the released version and set up the intended deployment path. Further development and maintenance are planned with clear responsibilities.
Your results
A usable version with a clear handover.
The agreed scope of delivery includes the developed function and the documentation necessary for its use and maintenance.
Approved software version
The agreed functions as a testable release. Fulfilled criteria, known limitations, and still open requirements are clearly documented.
Test and Model Documentation
Test cases, results, and used model versions. You can see with which data the function was evaluated and where additional tests may be required.
Technical handover
Documented components, interfaces, and provisioning. Usage rights, access to source code, and future maintenance are explicitly regulated in the project contract.
Platforms in the project
Which platform fits your task?
The development can complement SynaEYE, SynaHUB, or Vaidio. We first examine which functions and extension points the chosen platform actually provides.
Operational video analyticsContinuous AI video analytics for live camera streams
SynaEYE continuously analyses existing cameras in real time and combines detection, event logic, alerts, tracking and forensics in a European edge AI platform.
The right choice if you:
want to monitor security, safety or operational processes continuously with your cameras and process detected events immediately.
Job-based Visual AIVisual AI for freely defined inspection tasks
SynaHUB turns a visual question into an automated AI job. A camera, API or schedule supplies the image, a VLM evaluates the situation and returns a structured result.
The right choice if you:
want to inspect conditions, references, quality or individual visual questions without developing a separate analytics module for every task.
Enterprise platformEnterprise AI Vision for particularly broad requirements
Vaidio is an established enterprise platform with an extensive catalogue of standardised video analytics, powerful forensic search and a broad integration ecosystem.
The right choice if you:
require a comprehensive standardised enterprise environment with a wide range of analysis, multi-site operation, and established VMS integrations.
You operate the processing according to your security, latency and privacy requirements on-premises, on edge systems, hybrid, or in a private cloud environment.
Compare platformsFrequently Asked Questions
Questions before the project starts.
Does everything have to be newly developed for your task?
No. We first check the configuration and existing platform functions. Individual development is used where it provides a concrete additional benefit for the required process.
When is separate model training required?
If existing models do not adequately recognise the relevant objects or states, training may be useful. For this, we need suitable data. For other tasks, a configuration or a structured VLM check is sufficient.
How do your employees participate?
You describe the actual task and review intermediate results. Especially helpful are typical examples, difficult edge cases, and feedback from the people who will later work with the application.
How are new features accepted?
Before development, we agree on verifiable criteria. Acceptance is based on software tests, suitable image data, and expert review. Known limitations are documented.
Do you get access to the source code?
This will be specified in the project contract. Individual development, platform components, and external libraries may be subject to different licensing terms. Rights and the scope of delivery will be clarified before commissioning.
Can the application be further developed later?
Yes. Extensions are prioritised based on new requirements and implemented as further versions. Ongoing maintenance and support can additionally be agreed upon.
Next step
What function is missing from your solution?
Describe to us the desired recognition or workflow. We clarify which part should be configured and which should be specifically developed.
