Integrated Circuit Packaging Recognition with Tilt Auto Adjustment using Deep Learning Approach

Abstract

A deep-learning-based approach for recognizing integrated circuit (IC) packaging type is presented in this paper. The objective of this work is to design a deep-learning method that can recognize multiple types of packaging per detection, performing counting operations, and calculating the centre location of an IC with its tilting angle. The transfer learning from model You-Only-Look-Once (YOLO) v5 was chosen because it has been trained with the coco dataset and has a more reliable feature extraction system than the other models. In order to extract data from images, OpenCV was used, which allows the deep learning model to perform more efficient analysis of the input data. Apart from that, the principal component analysis (PCA) was used to estimate the angle of the IC in order to determine the rotation of each IC for the purpose of tilting adjustment. The developed model has an average confidence score of 85% and is capable of operating in a variety of conditions, as demonstrated by ANOVA analysis.

Keywords

• artificial intelligence
• deep learning
• integrated circuit packaging type
• image recognition
• anova analysis