Features 01
Expected Problems
?Don't you want to automate character recognition tasks performed in QA and QC, which requires high work load.
?Aren't you bothered by the difficulty of recognizing blurred or slanted characters?
?Don't you have trouble identifying characters that have been deformed by winding?
Features 02
Technology highlight
Point (1): Performs annotation (*) by surrounding the character outline with a polygon, and constructs a trained object detection model.
Point (2) Creates and applies a model with a function for selecting the optimum angle based on the reliability of AI detection, even for slanted character.
Character recognition with high accuracy is possible by combining technologies above, even characters are deformed, blurred, or smudged.
( * ): Task of assigning meaning to data.
Features 03
Proposed Solutions (Examples)
When product information is printed on a packaged product, the characters may be deformed or slanted due to changes in the cargo condition. Character recognition by manual often leads to human errors. It is possible to read characters automatically with higher accuracy than manual.
This technology is applicable to customers who perform product identification control using information printed on products.
Search for
