
New Features of Version 4.4
Ready for a new experience?
Shortened training & inference time
Fully Optimized for the
NVIDIA RTX 50 Series
With the release of the NVIDIA RTX 50 Series, we now provide optimized model support—bringing faster training, quicker inference, and enhanced AI performance with the latest hardware.
Users gain greater flexibility in building high-performance models across a wider range of hardware setups.

Fast and precise,
now for a broader range of models
Advanced AI Labeling Tools
Now supporting a broader range of models, our upgraded AI labeling tool delivers faster inference and more precise labeling.
-
Expanded Model Support
With just one click, the AI model automatically detects labeling areas, enabling faster and more precise labeling. -
Size Threshold Adjustment Added
Set size thresholds directly in the AI labeling tool to focus labeling on only the most relevant areas.

Create custom OCR models
tailored to your project
OCR Text Rule
The OCR Text Rule allows users to detect only the
text that matches specified formats and conditions
within images.
It improves model performance and reliability by
correcting recurring errors, such as misdetecting the number '0' as the letter 'O,' even after the model
has been trained.
.png)
Enhance inference speed
in multi-model environments
Enhanced Flowchart &
Inference Center
With this update, a single model can now be shared
across multiple classes, simplifying the structure of
multi-model workflow.
As the total number of required models is reduced,
the entire inspection flow becomes more streamlined,
resulting in faster real-time detection speeds.
.png)
Precise analysis of small defects
New Measurement Tools
New measurement tools have been added to
support precise inspections of micro defects.
-
Path Distance Measurement
Measure the length of various lines, particularly
useful for assessing curved distances. -
Detection Area Measurement
Measure the size and area of detected regions
for applications that require precise volume or
area analysis.
.png)