ReliVision consists of 4 fully integrated components.
Allows users with a complex and diversified portfolio of products to develop in-house AI models, without writing a single line of code. The models developed can run with ReliAudit.
Use the Node Editor for setting up ReliVision.
Use the Node Editor functionality of ReliVision to integrate the solution into the shop floor and manufacturing execution systems.
Plug any camera hardware and start capturing images of products or components from the Live View Screen.
Machine vision cameras, industrial grade cameras, smart cameras, static cameras, iPhone, and webcams are supported.
Select images to annotate and define specific regions that require inspection or quality monitoring.
Define regions of interest with a simple drag and drop interface.
From Relimetrics' library of deep learning models, select the one that best suits your use case and proceed with the training.
The dataset will be split for testing and validation.
All the progress can be monitored via the visualization dashboard.
Deploy the trained model on the shop floor and begin the quality audit.
Visualize, accept, or dispute the audit results using ReliUI.
Automates and digitizes visual inspections in-line, creating full traceability of quality in all stages of production.
Integrates with existing infrastructure on the factory floor and enhances existing QA solutions with advanced machine learning.
Inference can be done either at the edge or in the cloud.
ReliAudit works with industrial grade servers equipped with NVIDIA GPUs (e.g. T4, A2, A30)
Microsoft Azure, Google Cloud & AWS
ReliVision supports a wide library of industrial protocols and interfaces.
ReliVision can quickly integrate with shop floor systems using the NodeEditor.
Assembly issues flagged on the HMI can be disputed by an operator.
The information is filtered and used for retraining with ReliVision in-line retraining technology.
ReliVision provides well-defined deep learning recipes, enabling anyone to train deep learning models without AI or coding expertise.
Classifies an image.
Output: This image is an electronic circuit.
Detects and locates objects with rectangles.
Output: There are 4 electronic components in these positions.
Detects the exact count and shape of objects.
Output: There are 4 electronic components in these positions and these are their pixels.
ReliVision is hardware and image mode agnostic.
It can be integrated with any camera hardware and process any type of image data.