National University of Kaohsiung /  Prof. Tzung-Pei Hong

 Pain Points Solved 

This technology addresses the tediousness and high error rate faced by engineers in the Electronic Design Automation (EDA) field when organizing electronic component datasheet data. It proposes an automated identification and matching method centered on deep learning and text mining. 

 Technology Introduction 

Traditionally, engineers must page through datasheets from different manufacturers with varying formats to manually extract part numbers, packages, and other key information, which is time-consuming and prone to omissions or errors. This technology combines YOLOv5 image recognition and OCR text extraction to automatically identify headers, IC diagrams, tables, and text regions within datasheets. It utilizes pdfplumber to correct OCR recognition errors and further employs algorithms designed for different data types to automatically pair part numbers with packages. Experimental results show an accuracy rate of 97.99%, significantly reducing manual data organization time and error rates, while improving data digitization efficiency and the degree of design process automation.

▲Caption: Part number identification result on the YOLOv5 homepage (header).

▲Caption: Identification results of electronic component characteristic tables.

▲Caption: Text bounding box results for the IC diagram.

 Application Examples 

This technology is primarily applied to the digitization and organization of electronic component datasheet data, suitable for the workflow of Electronic Design Automation (EDA) engineers. Through automated identification and matching algorithms, it can rapidly extract part numbers, packages, tables, IC diagrams, and text information from datasheets of varying formats from different manufacturers, and automatically perform correct pairing, reducing manual organization time and error rates.

 Related Links 

None

 Patent Name and Number 

None

 Industry-Academia / Tech Transfer Partner 

None

 Honors and Awards  

None

 Technical Contact  

Vivian Lee, Administrative Assistant 

National University of Kaohsiung
Tel: +886 7-5916639
Email: vivianlee@nuk.edu.tw