NAVI AI PRO

Higher Performance, Lower Cost:
The New Standard for AI Vision Inspection

Contact Us

Get Started with AI Vision Inspection Without Experts.
Enjoy Incredible Performance Even on Low-End CPUs.

Fast AI Inference, Better Quality, Lower Costs:
All in One with NAVI AI PRO.

Features

  • 01

    No need to worry about costs when adopting AI!
    Adopt AI without replacing equipment!

    1ms Ultra-Fast AI Inference
    Using Only the CPU

    Achieve ultra-fast inference in less than 1 ms with a low-end CPU without a high-end GPU.

    Run high-performance AI models on low-end PCs, cutting infrastructure costs by over 90%, from tens of millions of dollars.
  • 02

    Is your model underperforming?
    Maybe it's the data.

    Eliminate Data Outliers and
    Imbalances

    Understand your data before the model.
    Pre-filter outliers and unbalanced data before training to create a highly reliable dataset.
    Maximize model performance through significant improvements in data quality.
  • 03

    Easy to Use for First-Time Users?

    Simple for Beginners,
    Precise for Experts

    It offers a personalized experience based on the user’s skill level, making it easy for AI beginners and experienced engineers alike.
    Run AI Training with a Single Click for Non-Technical Users
    Experts Can Configure Fine-Grained Under- and Over-Checking Conditions
  • 04

    Feel Free to Ask Questions!

    Your Own AI Coach:
    AI Advisor

    The industry's first generative AI coach analyzes your learning process and results, providing easy-to-understand explanations for every key step.
    Even first-time users can navigate easily and boost model performance.
  • 05

    Set Up Once,
    Say Goodbye to Relearning!

    Easy Retraining in Just
    3 Clicks

    It is inefficient to retrain the model from scratch every time a new defect type is introduced.

    Now you can quickly update your models to reflect new data in just three clicks and continue to improve performance.
  • 06

    Choose the Specialized Models
    You Need for Each Type!

    Optimized Models Reflecting
    the Characteristics of
    Each Process

    We offer domain-specific models for a variety of industries, including PCB, display, food and logistics, etc.
    Choose only the right model for your process and use it efficiently.
    MERCURY: Pass/fail judgment and type classification
    MARS: Unstructured area detection and classification
    VENUS: Object detection, classification, and dimension measurement
    JUPITER: Detection using only good product data

Higher Performance, Easier to Use
Achieve Perfect Efficiency with Automated AI Training

Functions

  • AI Advisor

    GPT for end-to-end model training

  • Data Analysis

    Minimize human error with proactive data analysis

  • Continual Learning

    Simple incremental training completed in three clicks

  • Auto Labeling

    Automatic labeling and key data selection using trained models

  • Auto Param

    Optimal hyperparameter settings within five minutes

  • User Mode

    Model training accessible even for non-experts

Success Stories Built
on Extensive On-Site Experience with Our Clients

Applications

See where our solutions work!

  • PCB

    Precise grading based on defect size and location (Good / Defective / AOR)
    A missed defect rate of 0.001%

  • Display

    Full quality checks for atypical defect types (Mura, Bit, Dot, etc.)
    Minimize errors with precision patterns in micro-defects such as surface cracks and delamination

  • Battery

    Detection/alert for falls, collapses, and lack of safety gear
    Smoke/fire detection and alert
    Real-time monitoring based on safety regulation manuals

  • Steel

    Robust against diffuse reflections
    Detect microscopic cracks that are invisible to the naked eye

FAQ

What are the minimum computer requirements to run NAVI AI PRO?

OS: Window 10 version, GPU: RTX 3060 8 GB, CPU: Intel i5-7700, memory DDR4 16 GB or more.

What is the minimum number of data to train a model?

You can get by with as few as 10 chapters, but we recommend around 100 chapters per class to ensure stable performance.

Can I use images from multiple cameras together?

Yes, NAVI AI Pro supports multi-channel training and can process images collected from different cameras simultaneously.

Can I use self-labeled images?

Yes, NAVI AI Pro supports not only NAVI formats, but also PASCAL VOC and COCO formats.

What CPU specifications are required for inference using only the CPU?

Any general-purpose x86-64 CPU capable of running Windows 10 or later can be used without restriction.

Can I configure my system with just the CPU and no GPU?

Training: A GPU is required for model training.
Inference: Field-processing machines can be configured with CPU-only operation for high-speed inference under 1 ms.

High-Performance Vision AI Solution
Start Now Easily Without Any Hassle

Contact Us Now