Contact Us

Improve Visual Inspections &
Achieve High-Precision

AI Software

The AI developed by HACARUS undergoes iterative learning from good data samples, enabling automation of visual inspection with the same level of accuracy as experienced inspectors.

Examples of Targeted Work and Defect Detection

Use Cases

Surface Inspections

Cosmetic Defects Such as Scratches, Dents and Contaminations

Completeness Check

Foreign Object Contamination in Electronic Circuit Boards, Soldering Defects, and more

Examples of Defect Detection by Inspection Algorithms

Scratches on Metal Plates

Dents on Metal Workpieces

Foreign Object Contamination on Electronic Circuit Boards

Are you facing challenges
like these?

Wanting to automate visual inspection but haven't been able to do so

  • Challanges in collecting and labeling defect in large quantities
  • Unable to conduct extensive re-training every time the workpiece changes

Using rule-based systems to implement inspection equipment

  • Already have inspection equipment
  • You want to prepare the imaging equipment on your own
  • Automated inspections take time to get up and running

AI has been introduced but not effectively utilized

  • AI in use but there are accuracy issues
  • AI (Deep Learning) has not been validated and cannot be applied to mass production.
  • Defects missed due to unknown anomalies


AI Software

An AI visual inspection software producing highly accurate predictions, trained with small data & only with good samples - at high speed. It does not require individual development for each inspection target or the use of GPU, running on-premise for both training and predictions.

Create an AI model from dozens of good samples data
An AI that can be continuously trained in short periods of time and with high frequency
Unlimited model creation within the license period

Until now

Each Inspection Target Requires Development of a New Inspection Model

Each inspection target needs its own AI model, which requires extensive development work.

After Introduction

No development required for each inspection target

AI Visual Inspection Models are created by iteratively learning from Small Data.

Comparison with other technologies

Model Training
with Good Data
Training Data
GPU Repeated
Visual Inspection
Rule Base
Visual Inspections
× ×
Deep Learning
Visual Inspections
Large A Few Hours Required ×
HACARUS Small A Few Minutes Unnecessary

Feel free to test out training and inspection using our HACARUS Check AI Software

Learn more

Case Study


30% Reduction in Time Spent On Visual Inspections

Secondary inspection of automotive precision parts (gears), with the goal of reducing defects missed and inspection man-hours in the existing inspection system.
(Training Data: 20 Good Goods)

Learning Time ~ 25 Seconds
Inference (Inspection) Time ~25 Msec
Detection Rate 97.8%

HACARUS can provide a one-stop solution from AI software to hardware. We can propose the necessary cameras, lighting, and equipment for image inspection according to your needs. Please feel free to consult with us.

Contact Us