Build, deploy and manage AI inspection solutions for industrial X-ray images from welds to full assemblies.
Detect critical defects (e.g., porosity, cracks, inclusions)
Build and deploy with your expertise (no coding)
Keep your data private and on-premise
X-Ray Nondestructive Testing (NDT)
The ReliVision platform empowers inspectors and NDT professionals to build, deploy and manage end-to-end AI inspection solutions without coding expertise. Leveraging their domain knowledge, users can automatically detect, classify and localize defects in industrial X-ray images, including corrosion, porosity, cracks and other material discontinuities. All data and AI models remain private and on-premise, ensuring full ownership and control over inspection processes.
Beyond trained AI models, ReliVision enables inspectors to create custom image-processing steps using natural language and reference images without writing code. Simply describe the inspection task, attach an example image, and ReliVision automatically generates the code. Here, this capability is demonstrated through porosity detection in pipeline weld X-rays, a critical quality-control application across oil and gas, chemical, power and water infrastructure.
Porosity in the weld can reduce weld strength and pressure integrity.
Automated radiograph analysis detects and quantifies porosity quickly and consistently.
Describe the task in plain language and attach a reference image. ReliVision generates the custom detection node for you.
Step 1
Describe what to detect and upload a reference image so the model understands the expected output, for example: “Annotate the white spots/blobs similar to the annotation in the attached image.” ReliVision generates a custom node that detects likely porosity candidates.
.jpg)
Step 2
Flag low-confidence or missed detections and improve the node instantly with extra guidance and sample images, for example by increasing detection sensitivity and including small and medium-sized porosity clusters. Every edit is kept in the node's history.
.jpg)
Result
The generated classical image-processing node segments porosity across the weld X-ray set, with each detected pore marked for review.
Read the NLP custom node guide.jpg)
See how an NDT team uses ReliVision to train, deploy and scale an AI model that detects and localizes corrosion in X-ray images.
Step 1
Import X-ray images into ReliUI, define corrosion states and label affected regions with the polygon tool. Create high-quality annotation sets directly on the platform, with no external tools or cloud access. Collaborate with your team and manage annotation sessions with full version control.

Step 2
Train corrosion-specific semantic segmentation models in a few steps. Select your dataset, configure the training settings and launch. ReliVision handles preprocessing, data splitting and model optimization, locally or on distributed compute.

Step 3
Build AI pipelines with your trained models in the visual Pipeline Editor. Upload image sets, configure segmentation blocks and run predictions instantly. Visualize corrosion zones with pixel-level overlays, all in a single interface.

Step 4
Deploy inspection pipelines to ReliAudit for real-time edge inference, and review audit results and performance metrics in ReliWeb with complete traceability. All modules run on-premise with no internet dependency, supporting industrial-grade data security.

Step 5
Handle multiple tasks with effective queuing, replicate high-performing solutions to more lines and sites at no cost, add hardware and software incrementally as needed, and scale up computing power seamlessly, all through ReliVision's synchronized distributed architecture.

Why ReliVision
NDT professionals create, deploy, and update AI-powered X-ray inspection solutions without programming expertise.
ReliVision detects and classifies corrosion, cracks, porosity, and other defects across varying inspection conditions.
NDT teams deploy and share images and AI models across multiple inspection stations while keeping data on-premise.
Before vs. with ReliVision
Before
Limited adaptability
Pre-trained AI models struggle to maintain accuracy as products, defects and inspection conditions change.
Complex and costly to maintain
Deploying, updating and maintaining AI applications requires specialized technical expertise.
Difficult to scale
Standalone inspection systems limit the sharing of data, models and improvements across production lines and plants.
With ReliVision
Accurate and adaptable
AI models detect defects across varying production conditions and can be easily updated as inspection requirements evolve.
Easy to build and maintain
Production teams create, deploy and update AI inspection applications without programming expertise.
Designed to scale
Teams share AI models, images and annotations across inspection stations and plants while keeping data private and on-premise.
The industry's only self-install, all-in-one-device industrial automation software for digitizing quality assurance and NDT.
