Neurocle Neuro-R Runtime Library

Highlights

Neurocle Neuro-R is a runtime library for deploying trained deep learning vision models into production inspection systems.

Models developed with Neuro-T or Neuro-T Engine can be integrated into machine vision applications for real-time inference across CPU, GPU, NPU, and embedded processing platforms.

From Trained Model to Production Inspection

STEP 1

Train

Develop and validate the inspection model in Neuro-T or Neuro-T Engine.

→

STEP 2

Integrate

Use Neuro-R APIs to connect the trained model with the machine vision application.

→

STEP 3

Inspect

Run the trained model against live production images or video streams in real time.

Flexible Model Deployment

Production inspection systems have different requirements for speed, processing power, enclosure space, and power consumption. Neuro-R allows the runtime hardware to be selected around the application instead of forcing every inspection system onto the same computing platform.

This gives machine builders and system integrators flexibility to deploy Neurocle models on conventional industrial PCs, GPU workstations, AI accelerators, or compact embedded platforms.

Supported Processing Platforms

GPU

CPU

NPU

Embedded Board

Select the processing architecture based on the application’s throughput, cost, power, and mechanical requirements.

Neurocle Neuro-R supported processing platforms

Programming Languages

C++

Neuro-R API integration

Python

Neuro-R API integration

Features

Model Optimization for High-Speed Inspection

Optimize trained models and processor performance around the inference speed required by high-throughput production inspection systems.

Neurocle Neuro-R model optimization for high-speed inspection

Model Optimization for Edge Devices

Use quantization and embedded-device optimization to create lightweight deployment models for systems with limited processing power, space, or power availability.

Neurocle Neuro-R edge device model optimization