基于主动学习和U-Net++分割的芯片封装空洞率的研究

doi:10.11896/jsjkx.220200092

计算机科学 ›› 2023, Vol. 50 ›› Issue (6A): 220200092-6.doi: 10.11896/jsjkx.220200092

• 图像处理&多媒体技术 • 上一篇下一篇

基于主动学习和U-Net++分割的芯片封装空洞率的研究

齐选龙¹, 陈弘扬², 赵文兵¹, 赵地^3,4, 高敬阳²

1 北京工业大学北京 100124;
2 北京化工大学北京 100029;
3 中国科学院计算技术研究所北京 100080;
4 中国科学院大学北京 100049

出版日期:2023-06-10 发布日期:2023-06-12
通讯作者: 赵文兵(zhaowb@bjut.edu.cn)
作者简介:(victorqi0712@gmail.com)
基金资助:
中国科学院计算技术研究所-华为联合实验室项目(YBN2020055088)

Study on BGA Packaging Void Rate Detection Based on Active Learning and U-Net++ Segmentation

QI Xuanlong¹, CHEN Hongyang², ZHAO Wenbing¹, ZHAO Di^3,4, GAO Jingyang²

1 Beijing University of Technology,Beijing 100124,China;
2 Beijing University of Chemical Technology,Beijing 100029,China;
3 Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100080,China;
4 Chinese Academy of Sciences,Beijing 100049,China

Online:2023-06-10 Published:2023-06-12
About author:QI Xuanlong,born in 2000,undergra-duate.His main research interests include deep learning,computer vision and so on. ZHAO Wenbing,born in 1973,Ph.D,lecturer.Her main research interests include deep learning and medical image processing.
Supported by:
Institute of Computing Technology, Chinese Academy of Sciences-Huawei Joint Laboratory Project(YBN2020055088).

摘要/Abstract

摘要： 内焊球空洞是BGA封装芯片的主要缺陷,可能会导致电气故障。目前,常用的检测方法是人工对照芯片X光影像检查,此类方法检测准确率低且时间、人力资源消耗大。因此,基于深度学习的自动化芯片缺陷检测方法越来越受到关注。芯片空洞检测与语义分割任务对应,但受限于数据缺乏高质量标注,模型准确率通常偏低,主动学习框架是潜在的解决方案。文中基于主动学习和U-Net++构建了芯片空洞率检测模型,通过等距划分将数据集分为多个子集,每个子集采用训练-预测-标注-扩展的框架循环优化U-Net++模型。在BGA封装芯片数据集上进行实验,模型分割平均Dice系数达到了80.99%,总体准确率达到了94.89%,达到了预定目标。首次将主动学习引入芯片检测领域,经实验验证可以有效提升芯片数据的标注水平,使得模型的分割准确率有所提高。

关键词: 内焊球空洞, 主动学习, 图像分割, 目标检测

Abstract: Bump void is one of the most common physical defects in BGA packaging,which may lead to electrical failures and shortened lifetime.At present,the commonly used quality inspection is based on manual check on X-ray images,which has low accuracy and high time consumption.Therefore,automated chip detection methods based on deep learning draws increasing attention in industry.This paper proposes an active learning and U-Net+＋based void rate detection network.Based on active lear-ning,we apply equidistant partition for the whole dataset.For each sub-dataset,we take training-prediction-labeling-extension as pattern to optimize U-Net+＋network.The average dice coefficient on separated model sets reaches 80.99% on test set,while the overall accuracy rate reaches 94.89%.We innovatively apply active learning in in-line defect detection,and the result shows that,it can effectively enhance the labeling standard of data and model’s division precision.

Key words: Bump void, Active learning, Image segmentation, Object Detection

中图分类号:

TP399

齐选龙, 陈弘扬, 赵文兵, 赵地, 高敬阳. 基于主动学习和U-Net++分割的芯片封装空洞率的研究[J]. 计算机科学, 2023, 50(6A): 220200092-6. https://doi.org/10.11896/jsjkx.220200092

QI Xuanlong, CHEN Hongyang, ZHAO Wenbing, ZHAO Di, GAO Jingyang. Study on BGA Packaging Void Rate Detection Based on Active Learning and U-Net++ Segmentation[J]. Computer Science, 2023, 50(6A): 220200092-6. https://doi.org/10.11896/jsjkx.220200092

参考文献

[1]YUNUS M,SRIHARI K,PITARRESI J M,et al.Effect ofvoids on the reliability of BGA/CSP solder joints[J].Microelectronics Reliability,2003,43(12):2077-2086.
[2]CHIU T C,ZENG K,STIERMAN R,et al.Effect of thermal aging on board level drop reliability for Pb-free BGA packages[C]//54th Electronic Components and Technology Conference.2004.
[3]YANG T,ZHANG S S,JIANG F Z,et al.Brachial plexus ultra-sound image optimization based on deep learning and adaptive contrast enhancement[J].Computer Science,2019,46(11A):236-240.
[4]PETERSON B,KWAN M,DUEWER F,et al.Optimizing x-ray inspection for advanaced packaging applications[C]//2020 International Wafer Level Packaging Conference.2020.
[5]CHAO F,XIAOMINL,KOW J.3D package failure analysischallenge and solution[C]//2015 IEEE 17th Electronics Packaging and Technology Conference.2015.
[6]VAGA R,BRYANT K.Recent advances in X-ray technology[C]//2016 Pan Pacific Microelectronics Symposium.2016.
[7]GONG J,UME I C.Void inspection in lead-free solder bumps on ball grid array(BGA) packages using laser ultrasound technique[C]//ASME International Mechanical Engineering Congress and Exposition.2011.
[8]LV H M,ZHAO D,CHI X B. Deep learning for early diagnosis ofalzheimer’s disease based on intensive AlexNet[J].Computer Science,2017,44(Z6):50-60.
[9]CHEN S W,LIU Y J,LIU D,et al.AlexNet model and adaptivecontrast enhancement based ultrasound Image classification[J].Computer Science,2019,46(6A):146-152.
[10]TABERNIK D,ELA S,SKVARC J,et al.Segmentation-based deep-learning approach for surface-defect detection[J].Journal of Intelligent Manufacturing,2020,31(3):759-776.
[11]ZHANG Z Z,GAO J Y,LV G,et al.Pathological image classification of gastric cancer based on depth learning[J].Computer Science,2018,45(11A):263-268.
[12]SAID A F,BENNETT B L,KARAM L J,et al.Automated void detection in solder balls in the presence of vias and other artifacts[J].IEEE Transactions on Components,Packaging and Manufacturing Technology,2012,2(11):1890-1901.
[13]PENG S,NAM H D.Void defect detection in ball grid arrayX-ray images using a new blob filter[J].Journal of Zhejiang University Science C,2012,13(11):840-849.
[14]WANG H,LIU D.A new threshold segmentation algorithm for segmenting micro-focus X-ray BGA solder joint image[C]//4th International Conference on Information Technology and Ma-nagement Innovation.2015.
[15]VAN V M.Void detection in solder bumps with deep learning[J].Microelectronics Reliability,2018,88:315-320.
[16]WANG F,WANG F.Rapidly void detection in TSVs with 2-D X-ray imaging and artificial neural networks[J].IEEE Transactions on Semiconductor Manufacturing,2014,27(2):246-251.
[17]HE K,GKIOXARI G,DOLLAR P,et al.Mask r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision.2017.
[18]NEELURU V K,AHUJA V.Void region segmentation in ball grid array using u-net approach and synthetic data[J].arXiv:1907.04222,2019.
[19]SCHIELE T,JANSCHE A,BERNTHALER T,et al.Comparison of deep learning-based image segmentation methods for the detection of voids in X-ray images of microelectronic components[C]//2021 IEEE 17th International Conference on Automation Science and Engineering.2021.
[20]SETTLES B.Active learning literature survey[R].University of Wisconsin-Madison,2009.
[21]ZHOU Z,RAHMAN S M M,TAJBAKHSH N,et al.UNet++:A nested u-net architecture for medical image segmentation[M]//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support.Cham:Springer,2018.
[22]RONNEBERGER O,FISCHER P,BROX T.U-Net:Convolu-tional networks for biomedical image segmentation[C]//International Conference on Medical Image Computing and Computer-assisted Intervention.Cham:Springer,2015.
[23]SU H Q,FU J C,GU H,et al.Parallel algorithm design for assisted diagnosis of prostate cancer[J].Computer Science,2019,46(11A):524-527.
[24]GAO Q,GAO J Y,ZHAO D.GNNI U-net:precise segmentation neural network of left ventricular contours for MRI image based on group normalization and nearest interpolation[J].Computer Science,2020,47(8):213-220.
[25]IGLOVIKOV V,SHVETS A.Ternausnet:U-net with vgg11encoder pre-trained on imagenet for image segmentation[J].arXiv:1801.05746,2018.
[26]TAN M,LE Q.Efficientnet:Rethinking model scaling for convolutional neural networks[C]//International Conference on Machine Learning.2019.
[27]KAMBLE R,SAMANTA P,SINGHAL N.Optic disc,cup and fovea detection from retinal images using U-Net+＋with EfficientNet encoder[C]//International Workshop on Ophthalmic Medical Image Analysis.Cham:Springer,2020.
[28]SONG Q,LI S,BAI Q,et al.Object detection method for gras-ping robot based on improved YOLOv5[J].Micromachines,2021,12(11):1273.
[29]BOCHKOVSKIY A,WANG C Y,LIAO H Y M.Yolov4:Optimal speed and accuracy of object detection[J].arXiv:2004.10934,2020.
[30]REZATOFIGHI H,TSOI N,GWAK J Y,et al.Generalized intersection over union:A metric and a loss for bounding box regression[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019.
[31]CHEN H Y,GAO J Y,ZHAO D,et al.LFSCA-UNet:liver fibrosis region segmentation network based on spatial and channel attention mechanisms[J].Journal of Image and Graphics,2021,26(9):2121-2134.
[32]CHEN H Y,GAO J Y,ZHAO D,et al.Review of the research progress in deep learning and biomedical image analysis till 2020[J].Journal of Image of Graphics,2021,26(3):475-486.
[33]REN P,XIAO Y,CHANG X,et al.A survey of deep activelearning[J].ACM Computing Surveys,2021,54(9):1-40.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于主动学习和U-Net++分割的芯片封装空洞率的研究

Study on BGA Packaging Void Rate Detection Based on Active Learning and U-Net++ Segmentation

PDF (PC)

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 0

Metrics

本文评价

推荐阅读 0