BUSINESS

NAVI AI Software

The NAVI AI Software utilizes deep learning algorithms to learn and analyze images, allowing for defects inspection that are difficult to detect with traditional vision inspection methods. In addition to powerful AI models for learning and inference, NAVI AI Software provides end-to-end technology that can bring you higher performance at a lower cost in developing AI projects.

Why LAON PEOPLE?
Even if you are not an AI expert, NAVI AI provides guides from the data gathering process to deriving learning outcomes process. You can implement the best project under the guidance of experts, including how to collect data to increase its utilization, how to overcome the limitation of small data, and how to maximize the effectiveness of learning.
INQUIRY
sales@laonpeople.com+82-1899-3058ONLINE INQUIRY

Product Specific Specifications

Information
SCROLL >>

 

NAVI AI PRO

Minimize development period until
deployment stage with Data-centric AI

 









AI software can no longer derive good results just by providing good AI models.
Existing AI companies focused on just the development of AI models, field applications, and the cost of AI models.
They postponed data preparation and refinement to customers who have domain knowledge.

As a result, the overall AI development period takes too long and the project's success rate was getting lower.

Features of industrial AI projects

Small data sets and short term
Most of the projects need to be completed in a short period of time with a small data set.
Unbalance of data and deviation of period for gathering data
Too much major defects, but no intermittent defects
Difficulty in data recycling
Difficult to recycle data when conditions have changed.


LaonPeople can do.

LaonPeople's NAVI AI PRO provides various data management and
analysis functions for AI that can reduce the development period and field application dramatically.







주요 특징

NAVI AI PROimproved customer convenience greatly
and increased the success rate of the entire AI project significantly, by equipping tools that make the process of preparing and refining data easier and faster.

Document-based metadata database management

* Facilitates history management by creating and recording various additional information to data.
* Increase the reusability of data through history management from gathered data

Efficient Learning Methodology

* Much faster learning can be performed in the data gathering process by analyzing the confidence of the data learned by AI.
* Images can be selected easily that need additional training or evaluation on a large amount of data

AI model with higher performance

* Perfectly optimized AI model for manufacturing industry by equipping latest research algorithms
* AI model that can be applied in various customer environments.

Easy Interface

* Reduced data set building time with convenient tool interface and functions
* Easy and convenient development environment.
* Benchmarking and comparison testing with various reference models with various performance evaluations are available






Higher accuracy with
less data and
shorter development period.



Product name Functions Toolkit Library
Mercury Mars Venus Jupiter Trainer Converter Library
Description High-speed pass/fail determination algorithm using AI algorithm (Classification) High-speed error detection algorithm using AI algorithm (Segmentation) High-speed object detection algorithm using AI algorithm (Detection) High-speed status determination algorithm using AI algorithm (Unsupervised) A tool that provides learn and performance test functions with easy and intuitive UI A tool to optimize the trained model using NAVI AI Trainer Provides API in C++, C#, and python environments so that users can develop NAVI AI tool according to the environment they want.
Feature * Type classification of images
* Multi-channel learning/inspection
* 500Mpixel/sec high-speed AI inspection per GPU
* Detection of small-scale defects in images
* Multi-channel learning/inspection
* 1000Mpixel/sec ultra-high speed AI inspection per GPU
* Simultaneous implementation of object detection, position determination and type determination
* Constant speed maintenance (regardless of number of detections)
* Excellent detection performance (regardless of rotation, size, or angle change)
* Defect detection using fair product images
* Unsupervised learning method
* Supports various resolution images
* Data set management using metadata
* Provides an intuitive UI in familiar Windows environment
* Supports large number of files.
* Provide supervised/unsupervised learning and inspection methods
* Optimize FP32/FP16 using TensorRT in environments with GPU
* Optimize FP32/FP16 using OpenVINO in environment without GPU.
* Provides API in C++, C#, and python environments
* Provides detection result with blob, probability map image, etc. (intuitive interface
* Can be operated under various conditions (GPU, CPU, etc.)
Application * Classification of objects, determination of pass or fail in quality, determination of types of defects * Detection of defective areas, detection of outliers * Object classification, position determination, counting * Detection of defective areas, detection of outliers * Can be applied in all industries that need to inspect defects * When it is necessary to optimize PC for faster speed * When it is necessary to develop in specific UI structure

Functions

 
Overview
What is the NAVI AI, vision inspection software using artificial intelligence?

● It overcame the limits of vision inspection with deep learning algorithm.
● It makes vision inspections more convenient by the machine learning-based study.
● It's the most accurate and convenient OCR reading in the world.





Screen Composition of Vision Inspection Software

Menu & Toolbar
The menu is a hierarchical structure that allows you to conveniently perform all vision inspections, consisting of Mode / File / View / Inspection / Tool / Window / Help.
And the function frequently used is provide as the form of icon in toolbars.

Display View
It shows imported images (or camera images) and real-time results set by the tool.

Tutorial
It provides easy learning to make a master of vision by using a project with tutorial representation.

Inspection Tool
Clients can manufacture any kind of vision inspection that clients want by using powerful function that 20 deep learning-based tools (NAVI AI Mercury / Venus / Mars) and computer vision-based tools.

Inspection Sequence Window
Applications built in the vision inspection tool can be sorted by inspection sequence, from a single vision inspection to a composite vision inspection, with a single vision software




Key Functions
Deep Learning Algorithm Tool
* NAVI AI Mercury
NAVI AI Mercury is available for high-speed accepted/defective judgment and high-speed object classification. 
NAVI AI Mercury runs at a much faster rate, omitting the location tracking algorithm.
How to set up NAVI AI Mercury for the vision inspection?
NAVI AI Mercury learns itself with a deep learning-based algorithm.
You only need to prepare pass/faulty images.

How does NAVI AI Mercury work?
<Learning Process>
User inputs pass/faulty learning data
NAVI AI creates hundreds to thousands of images and learns by itself

<Inspection Procedure>
User inputs the image for inspection
Output whether the product is pass or faulty





Overview  

What is the NAVI AI, vision inspection software using artificial intelligence?  
● It overcame the limits of vision inspection with deep learning algorithm.
● It makes vision inspections more convenient by the machine learning-based study.  

 



Screen Composition of Vision Inspection Software 
Menu & Toolbar
The menu is a hierarchical structure that allows you to conveniently perform all vision inspections, consisting of Mode / File / View / Inspection / Tool / Window / Help. And the function frequently used is provide as the form of icon in toolbars.
Display View
It shows imported images (or camera images) and real-time results set by the tool.

Tutorial
It provides easy learning to make a master of vision by using a project with tutorial representation.

Inspection Tool
Clients can manufacture any kind of vision inspection that clients want by using powerful function that 20 deep learning-based tools (NAVI AI Mercury / Venus / Mars) and computer vision-based tools.
Inspection Sequence Window
Applications built in the vision inspection tool can be sorted by inspection sequence, from a single vision inspection to a composite vision inspection, with a single vision software




 
Key Functions  

 Deep Learning Algorithm Tool

* NAVI AI Venus
NAVI AI Venus is applied to the pattern matching, object position detection, and counting. It applies deep learning and can detect through learning even in very complicated environments such as incomplete optical environment, bad focus, superposition of object, low light environment, complex background, and change of object scale.
How to set the NAVI AI Venus on the vision inspection?

What you have to do is to set the reference image or pattern.
NAVI AI Venus uses ImageGen ™ technology to create hundreds to tens of thousands of data for learning and learns how to detect even in the challenging environments.

How does NAVI AI Venus work?

<Learning Process>
User sets the inspection object.
NAVI AI learns by using deep learning.


<Inspection Procedure>
User inputs the image for inspection then, NAVI AI Venus detects and outputs the location of the inspection


     
Application Area 
  • Pattern Matching, Detection of Object Location, Counting

Overview

What is the NAVI AI, vision inspection software using artificial intelligence?
● It overcame the limits of vision inspection with deep learning algorithm.
●It makes vision inspections more convenient by the machine learning-based study.
● It's the most accurate and convenient OCR reading in the world.




Screen Composition of Vision Inspection Software
Menu & Toolbar
The menu is a hierarchical structure that allows you to conveniently perform all vision inspections, consisting of Mode / File / View / Inspection / Tool / Window / Help.
And the function frequently used is provide as the form of icon in toolbars.

Display View

It shows imported images (or camera images) and real-time results set by the tool.

Tutorial

It provides easy learning to make a master of vision by using a project with tutorial representation.
Inspection Tool
Clients can manufacture any kind of vision inspection that clients want by using powerful function that 20 deep learning-based tools (NAVI AI Mercury / Venus / Mars) and computer vision-based tools.
Inspection Sequence Window

Applications built in the vision inspection tool can be sorted by inspection sequence, from a single vision inspection to a composite vision inspection, with a single vision software.




Key Functions
■ Deep Learning Algorithm Tool

* NAVI AI Mars
NAVI AI Mars can be used in areas where vision inspection is tricky, such as atypical crack, defect, damage, and detection of stain spot. It solves the intricate problems including inspection on surface/exterior of metals, detection of the fine defect on solar panels, fabric inspection (Seam/Stain/Thickness/Cutout/Attachment), detection of defective parts, inspection on LCD sensitivities, and X-ray/CT/MRI/ultrasonic waves inspection for medical area. A deep learning-based semantic segmentation based is applied and accurately displays the defect spot on the screen.
How to set up NAVI AI Mars on the vision inspection?

Settings of NAVI AI Mars are similar to NAVI AI Mercury. 
It learns by itself with deep learning-based algorithm, and you need to prepare the accepted/defective images.
How does NAVI AI Mars work?
User inputs accepted/defective learning data.
NAVI AI learns after creating hundreds to thousands of images


User inputs image for inspection
NAVI AI Mars outputs whether the product is pass or faulty and displays the defect spot




Application Areas 
  • Atypical Crack, Defect, Damage, Stain
  • Fine Detection of Solar Panels and Inspection of LCD Sensitivities

Jupiter Tool 정보 확인하기

Introduction of Products

 

It is a high-speed error judgment Algorithm using AI.

 

 

Feature

  

  • ​Unsupervised Learning, which is learned by only fair quality, possible to detect errors.
  • Ultra-high-Speed AI Inspection (100M Pixel/Sec each GPU)






Future Value or Applicable Area

  • Detection of fail/pass quality of the product
  • Inspection of error in the print/OCR

 

Toolkit


Features

Intuitive UI for familiar window environments

Processing tens and thousands of large files within a short time

Offering supervised/unsupervised learning and inspection methods

 

Overview 

  AI Vision Inspection Image Learning

 

NAVI AI Trainer is a learning tool for AI Vision Inspection.

NAVI AI tool (Mercury, Mars, Venus) provides the function to perform learning and performance tests in a very easy and intuitive UI. The tool also allows users to learn and apply images directly to the inspection to reduce the amount of time that occurs when the subject to inspection changes frequently or when learning a new examination image. Of course, even if users do not have expertise in AI vision.






Main Features

* NAVI AI Mercury

It is a high-speed algorithm for judging accepted/defective products using AI algorithm. It accurately determines the minute differences and variations that are difficult to judge with rule-based algorithm. It provides targeted parts’ classification by deep learning. The procedure is explained below. 

Load the image to learn.

Set the classification class for each image.

When the class set of all images are completed, press the 'Training' button.

Run the generated .dat file through NAVI AI Mercury and proceed with the inspection.







* NAVI AI Venus

It is a high-speed Multiple object detector using AI algorithm. It allows robust detection of changes in size, position, and shape that are difficult to respond with pattern matching.

Learning objects are managed by multi-class territories. And with learning a selected territory, it detects locations and quantity of objects.

The procedure is explained below.


Load images that contain objects to be inspected.

Display the part of the image that needs to be detected as a box.

When the box is displayed on the image, press the 'Training' button.

Run the generated .dat file through NAVI AI Venus and proceed with the inspection.

     






     

     

    * NAVI AI Mars

     It is a high-speed defect Segmentation detector using AI algorithm. It accurately detects complicated external defects such as an unusual crack, damage, and stains. The procedure is explained below.

       

           ● Classify the accepted/defective images.

           ● Mask the defective parts from the classified defective image.

           ● When the classification and masking of the images are completed, press the 'Training' button.

           ● Run the generated .dat file through NAVI AI Mars and proceed with the inspection.

     



     

    Auto Labeling Tool 정보 확인하기

    Introduction of Products

     

    It is a high-speed Auto labeling tool for the image using AI algorithm.

     

     

    Feature

     

    • Semi-Supervised learning Algorithm that learns the classified images and then detect the rest of the images.


    Future Value or Applicable Area

    • Using for data classification and labeling

     

    Library

    Overview

    A deep learning software library even general programmers can use, you can program what you want even if you do not have knowledge of deep learning!


    * What is NAVI AI Library?​

    ● The NAVI AI Library is the only deep learning library specialized to Machine Vision area in the world.

    ● Programmers who do not know deep learning can also create a deep learning inspection program with deep learning library.

    ● It is provided as a DLL-based library, can be easily ported to existing programs and available to produce the customized programs.

    ● It provides C ++/C#/Python Interface/sample code.

    ● It supports GPU(CUDA)/CPU mode.

     

     



    Data Flow & Architecture

    ● It is a systematic deep learning architecture.
    ● You can easily proceed with the vision inspection using the .dat file generated by the trainer.
    ● It provides artificial intelligence inspection as well as rule-based vision inspection.

     

     








    Composition
    * Training License




    * Inspection License



    * Identification License