V-01 / INSPECTION
AIThon
An industrial machine-vision platform for algorithm development and shared inference. Design algorithm workflows, annotate data, and train and evaluate four model types in one desktop application, then provide centrally managed inference services to independent business applications.
Back to family catalogue←AIThon: algorithms, data and models
Centrally managed algorithms and independent business systems
AIThon An industrial machine-vision platform that brings algorithm development, data annotation, model training, evaluation, export and inference-service management into one desktop application. It provides shared algorithms and computing resources to multiple business applications.
Business applications VisionCore, AssyCore and V2Verity run independently and use AIThon's analysis and inference capabilities. Each application handles shop-floor data acquisition, process control and result presentation.
Standard operators and visual orchestration
Standard vision operators
Operators cover a range of machine-vision functions and can be combined into workflows for routine inspection. Availability depends on the current software version.
Configure routine inspection
Build flows by dragging and connecting operators, then setting parameters, regions of interest (ROIs) and rules. Routine inspections do not require custom application code.
Data preparation and annotation tools
Import, annotation and review
Prepare training data in one environment with data import, video frame extraction, dataset splitting and quality statistics, plus manual, automatic and batch annotation and batch review.
Visual tasks covered
Data and algorithm tools cover classification, object detection, segmentation, anomaly detection, human detection and OCR. The four training-model types are listed separately below.
Four model types: from training to inference reuse
Select object detection, instance segmentation, anomaly detection or object classification and organise model training by project.
Evaluate accuracy and review results in the integrated environment, managing models, training results and evaluation results independently for each project.
Export the model and embed it directly into an AI inference operator in the standard algorithm designer, then provide inference to business systems through algorithm services.
AIThon: shared inference and business-system coordination
Business systems sharing AIThon capabilities
VisionCore | Appearance inspection
Manage equipment, optical systems, fields of view, ROIs and inspection plans on the desktop, alongside line integration, robot-arm and optical-device control, automatic inspection and history traceability.
AssyCore | Process inspection
Inspect compliance with manual SOPs, using a web interface to configure operation sequences and inspection rules for automatic inspection and alarms.
V2Verity | AI-assisted review
Connect to an existing AOI system through PLC or API interfaces, receive captured images and use AIThon for secondary review, followed by alarms, archiving and export. Existing optics can be reused when the imaging conditions are suitable.
Site configuration, asset reuse and project isolation
Shop-floor inputs and business outputs
Inputs include inspection signals from PLCs, images, video, datasets, ROIs, thresholds, recipes and business rules. Business applications display OK, NG and exception results, alarms and history. PLC, MES and reporting interfaces are adapted for each project.
Equipment, optics and interfaces
Associated business applications can combine cameras, lenses, lighting and robots from multiple manufacturers into several optical configurations. PLC, AOI and MES triggering, handshakes and result write-back are configured to match each protocol.
Data, model and algorithm assets
Retain annotated data, model records and algorithm templates as reusable assets. Validated data, models, algorithm chains and recipes can support repeated deployment at similar workstations.
Independent project management
Each business system stores its own business configuration and inspection history, reducing coupling between systems and projects while sharing AIThon's algorithm capabilities.
Start by defining the operational challenge.
Share the current process, constraints and intended outcome. We will assess the smallest verifiable starting scope.