Results during training, label check, defect image synthesis, reports and installation are improved.
Update · Software · IntLearn Pro · Version 2.2.0 · Posted November 15, 2026
IntLearn Pro is deep learning inspection software that finds defects like an expert. The AI learns the OK and NG calls that experts mark on photos and judges on the line by the same criteria, with 12 engines and modules, the most in Korea, in one program.
Main features
Six AI engines: anomaly detection (IntAnomaly), segmentation (IntSegmentation), OK/NG classification (IntClassification), detection (IntDetection), change comparison (IntComparison) and text recognition (IntOCR).
Six modules: defect synthesis (IntSynthesis), data augmentation (IntAugmentation), image enhancement (IntEnhancement), alignment (IntAlignment), label assist (IntLabelAssist) and measurement (IntMeasurement).
Direct it in plain words: tell the conversational AI (IntWise LLM) what to train and inspect in everyday words.
Inspection package: the model and its criteria go into one file (.izpkg) and are applied on the line.
What changed this time
IntLearn Pro 2.2.0 brings the following changes.
Results during training: while a detection model trains, the evaluation screen shows its detections right away. You can ask for an evaluation when you want one, without waiting for the save interval.
Label check: the label check button on the training status screen lets you look again at the labels used for training.
Defect image synthesis: it learns where labeled defects really occur and places synthesized defects there, then refines them with AI so they look natural.
Reports: the detection report shows the whole image, and a double click enlarges it to its original size.
History and recent projects: the detection, classification, OCR and anomaly detection screens have a work history tab, and recently used projects appear at the top.
Installation: one installer now includes OCR as well.
Fixed
Images could not be read when a folder or file name contained Korean characters. This is fixed.
Added recently
OCR: large images are split into tiles to find text. Full OCR has a progress bar and a stop button, and images can be rotated in 90° steps.
Project list: choose the sort order, and rename a project with the rename button.
Segmentation training charts: per-class accuracy (IoU) and the best-record points are shown on the charts.
Anomaly detection screen: the panels are rearranged in the order of work.
Display language: the main menu and the diagnostic dialogs follow the selected language.
When you need the update
Tell us through Contact Us at the top of the screen.