Availability and Procedure for Certification acc. to ASNT 9712 in Europe
ASNT 9712 is a global, independent, third-party certification scheme for NDT professionals that validates competence in accordance with the ISO 9712:2021.
ATG develops and delivers custom Automated Defect Recognition (ADR) systems for non-destructive testing. Our ADR solutions combine industrial machine vision, controlled illumination, automated part handling, NDT expertise and AI-based evaluation to support reliable inspection of serially produced parts.
ADR systems can be configured for selected NDT methods, including Visual Testing (VT), Fluorescent Magnetic Particle Testing (MT/MPI), and Fluorescent Penetrant Testing (PT/FPI). Depending on the application, systems can be delivered as standalone inspection stations or integrated into automated production lines with conveyors, robotic manipulation, Cartesian manipulators, and/or bin-picking.
A successful ADR project requires extensive cooperation between ATG and the customer. AI-based detection models are trained and validated on customer-specific parts and real inspection conditions. Thus, the provision of sufficient representative samples is essential. These samples should include typical OK parts, NOK parts, and parts with relevant indications encompassing expected production variability, surface conditions, geometries, material states, and borderline cases. Detection capability, classification accuracy, cycle time, and measurement precision are therefore determined individually for each application. They depend on the particular NDT method, part geometry, surface condition, illumination, camera configuration, indication type, acceptance criteria, the quality of training and validation data. Without representative training and validation samples, reliable ADR performance and the achievable level of automation cannot be assured.
Automated Defect Recognition (ADR) is an advanced approach to non-destructive testing in which image acquisition, indication detection, and evaluation support are performed automatically using cameras, sensors, controlled illumination, and AI-based algorithms.
ADR systems are designed to improve the consistency, repeatability, productivity and traceability of inspection workflows, particularly effectively in high-volume production environments with stable and well-defined inspection conditions.
ADR in NDT inspection can enable:
Automated image acquisition along with inspection support
Digital inspection records therefore traceability
Repeatable evaluation of candidate indications
Reduced operator workload in repetitive inspection tasks
Short cycle times while continuously operating
Configurable routing of OK, NOK, and uncertain parts
Statistical reporting as well as long-term process monitoring
24/7 operation
Visual Testing (VT) – ideal for detecting visible cracks
Magnetic Particle Inspection (MPI) – for surface and near-surface flaws
Fluorescent Penetrant Inspection (FPI) – effective for detecting surface defects under UV light
Ultrasonic Testing (UT) – used in spot weld inspection among other applications
ASNT 9712 is a global, independent, third-party certification scheme for NDT professionals that validates competence in accordance with the ISO 9712:2021.
ADR systems implemented in Non-Destructive Testing typically includes several interconnected steps:
Industrial cameras, optics, and customized lighting are selected according to the inspection task. The vision system is designed to capture required inspection areas defined in the customer-approved inspection procedure.
For suitable part geometries, the system can be configured to achieve full coverage of specified inspection surfaces. Actual coverage is verified during feasibility testing and validation.
The system can support OK/NOK decision workflows by applying validated customer-approved rules to inspection results. Depending on the application, uncertain results can be routed for human review.
Fully-automated sorting may be implemented only after customer-specific validation confirms that the system performance meets the agreed inspection procedure, acceptance criteria, production requirements, and existing regulations.
Automated Defect Recognition system implementation is a step-by-step process designed to reduce technical risk and verify performance under realistic conditions.
The process typically begins with a feasibility study. During this phase, the customer provides technical requirements, information about the inspected parts and representative OK/NOK samples or samples containing relevant indications.
Based on these inputs, ATG evaluates the suitability of the selected NDT method, tests part manipulation concepts, designs the vision-system configuration, and performs initial AI-based indication detection using customer-specific image data.
The objective of the feasibility study is to confirm whether the required indications can be reliably detected under controlled and repeatable conditions.
After a successful feasibility study, the next phase is often the implementation of a semi-automatic ADR system in the production environment.
In this stage, the system captures inspection data while automatically detecting and highlighting potential indications. Final evaluation and acceptance decisions may remain with qualified personnel while the system performance is being verified under real production conditions.
This approach supports safe validation while enabling additional data collection which allows AI models, thresholds, and inspection rules to be refined further before higher degrees of automation are introduced.
Once the ADR system has demonstrated stable performance according to agreed validation criteria on defined parts, it may be upgraded for automated OK/NOK sorting if permitted by existing regulations in a strictly-defined application.
Fully automated sorting is implemented only where the inspection process, acceptance criteria, system performance, traceability, handling of uncertain results, and fail-safe behavior have been validated and approved by the customer.
Even in automated systems, periodic validation, process supervision, maintenance, model-version control, and review of uncertain or exceptional cases remains important.
Each ADR system is developed and validated for a specific application. Performance depends on the quality and representativeness of training and validation data operating in stable production conditions with clearly defined acceptance criteria.
ATG can provide application-specific configurations and validation support, including:
Controlled image acquisition setup
Individualized AI model version and software configuration
Dataset preparation and annotation support
Detection testing
Definition of OK/NOK/uncertain categories
Traceable inspection records
Statistical reporting of inspection results
Controlled model retraining and update
Periodic performance review during production use
Model updates, retraining, threshold changes and software modifications should be managed through a controlled change process. This helps maintain stable inspection performance and supports auditability over the system lifecycle.
Customer samples, images and inspection data are used for feasibility testing, training or validation only under agreed contractual conditions. Data handling, retention, confidentiality and any use for model improvement are predefined with the customer before the project begins.
The complexity of an ADR project increases with part geometry, surface roughness, material variability, accessibility of inspected areas considering the complexity of relevant indications.
The most suitable candidates for ADR automation are usually serially produced non-critical parts with stable geometry, repeatable production conditions, clearly defined acceptance criteria, and sufficient inspection volume.
In suitable high-volume applications, customers may achieve return on investment within approximately 1 to 3 years, depending on production volume, labor costs, inspection scope, false-call handling, maintenance requirements, required cycle time and the level of automation.
A feasibility study is recommended to estimate technical complexity, expected performance and economic return for each specific application.
Depending on the application, these systems may incorporate:
ATG iVT (intelligent Visual Testing) systems support automated visual inspection of selected parts and surfaces. Depending on the application, iVT systems can detect and classify visible candidate indications such as scratches, corrosion, cracks, porosity-like surface features, pressure marks, burnouts, incorrect assembly, foreign objects, dimensional anomalies, or incorrect markings.
ATG iMT systems (intelligent Magnetic Testing) offer semi-automated or fully automated Magnetic Particle Inspection Applications.
These systems can assist with the assessment of the presence, type, position, and size of candidate indications. Automated MT inspection is technically demanding because results can be affected by part shape, surface condition, background fluorescence, magnetization conditions, illumination, and image-acquisition quality.
ATG addresses these challenges by combining NDT know-how, controlled inspection hardware, machine vision, and AI-based image evaluation. Each iMT system is configured and validated for the specified part and its inspection criteria.
ATG iPT systems (intelligent Penetrant Testing) support automated Fluorescent Penetrant Testing applications, including loading, unloading, image acquisition, indication detection, measurement, reporting, and sorting.
Important note: ADR performance is application-specific. Detection capability, classification accuracy, measurement precision and cycle time depend on the inspected part, NDT method, surface condition, optical configuration, dataset quality, and acceptance criteria. Each ADR system must be validated for its intended use before being used for production acceptance or automated sorting. Final responsibility for inspection procedures, acceptance criteria, and product release remains with the customer according to applicable standards, specifications, and regulatory requirements.
Automated Defect Recognition can improve the consistency, repeatability, productivity (24/7), and traceability of NDT inspection workflows.
ADR systems reduce the impact of operator fatigue in repetitive inspection tasks, support short cycle times, and enable digital storage of inspection data. This helps with quality audits, process optimization, defect-trend analysis, and long-term production monitoring.
ADR can also reduce dependency on manual visual evaluation in suitable high-volume applications, while qualified personnel remain responsible for procedure approval, supervision, and acceptance criteria.
ADR is usually introduced step by step.
In the first phase, the system acts as a decision-support tool by detecting and highlighting candidate indications while qualified personnel retain responsibility for final evaluation.
After sufficient production data, validation, and customer approval of indication detection rate, ADR may be used for automated sorting in defined applications. Even then, human oversight, periodic validation, handling of uncertain results, maintenance, control of software, and model changes remain essential.
Whether ADR can be used for automated acceptance depends on the customer’s inspection procedure, applicable standards, contractual requirements, product criticality and validation results. However, even in fully automated systems, human oversight remains important for system supervision, validation, and legal responsibility for the testing parts.
ADR systems are used in industries where quality, reliability, productivity and traceability are important. Typical applications include automotive, railway, defense, engineering, and other sectors with serial production, defined inspection requirements, and suitable regulations.
ADR is particularly suitable for high-volume production where parts, defect criteria, and inspection conditions are stable and repeatable.
ADR performance is influenced by several factors, including:
Quality and representativeness of training and validation data
Correct annotation of relevant indications
Selection of the appropriate NDT method
Stability of inspection conditions
Part geometry and surface condition
Illumination and camera configuration
Image quality and repeatability
Definition of acceptance criteria
Production variability over time
ADR can be applied to selected NDT methods where indication information can be captured as reliable image or sensor data. Common applications include:
Visual Testing (VT)
Magnetic Particle Testing (MT)
Penetrant Testing (PT)
Radiography Testing.
Each method requires a specific approach to illumination, image acquisition, and AI evaluation. The suitability of a method depends on defect type, part geometry, material, surface condition, production volume, and inspection criteria.
Yes, ADR can be applied to complex geometries. However, system complexity increases with part shape, surface roughness, accessibility of inspection areas and variability of surface appearance.
Complex applications may require robotic manipulation, multiple cameras, advanced lighting, part-specific fixtures, or optimized inspection paths.
A feasibility study is recommended to verify detection capability, surface coverage, cycle time and the optimal system configuration.
Implementation time depends on system complexity, part availability, required automation level, availability of representative samples, and required validation depth.
A feasibility study can often be completed within several months. Full implementation, including mechanical design, image acquisition, data collection, AI training, software integration, testing and production validation usually takes longer and has a project-specific timeline.
A stepwise implementation approach helps reduce technical risk and supports smooth integration into existing production lines.
ADR systems may achieve return on investment within approximately 1 to 3 years in suitable high-volume applications.
Actual ROI depends on inspection scope, labor cost, production volume, current inspection efficiency, required cycle time, maintenance costs, false-call handling, and the selected level of automation.
A feasibility study can help estimate both technical suitability and expected economic benefit.
Take a look at our non-destructive testing systems and equipment.