|本期目录/Table of Contents|

[1]何军红,温观发,黎长鑫.优化Faster R-CNN算法的小样本缺陷检测研究[J].工业仪表与自动化装置,2024,(05):94-101.[doi:DOI:10.19950/j.cnki.cn61-1121/th.2024.05.018]
　HE Junhong,WEN Guanfa,LI Changxin.Research on small sample defect detection based on optimized Faster R-CNN algorithm[J].Industrial Instrumentation & Automation,2024,(05):94-101.[doi:DOI:10.19950/j.cnki.cn61-1121/th.2024.05.018]
点击复制

优化Faster R-CNN算法的小样本缺陷检测研究(PDF)

《工业仪表与自动化装置》[ISSN:1000-0682/CN:61-1121/TH]

卷:
期数:
2024年05期
页码:
94-101
栏目:
出版日期:
2024-10-15

文章信息/Info

Title:
Research on small sample defect detection based on optimized Faster R-CNN algorithm
文章编号:
1000-0682(2024)05-0094-08
作者:
何军红; 温观发; 黎长鑫
(西北工业大学 航海学院，陕西 西安 710072)
Author(s):
HE Junhong; WEN Guanfa; LI Changxin
(School ofMarine Science and Technology，Northwestern Polytechnical University，Shaanxi Xi′an 710072，China )
关键词:
小样本; Faster R －CNN; 幻觉网络; 特征金字塔网络; 注意力机制; 缺陷检测
Keywords:
small －sample;  Faster R －CNN;  illusion network;  feature pyramid network;  attentionmechanism; defect detecting
分类号:
TP391.4
DOI:
DOI:10.19950/j.cnki.cn61-1121/th.2024.05.018
文献标志码:
A
摘要:
随着自动化检测技术的发展，基于深度学习的缺陷检测技术以其高精度、高效率、非接触性的特点正逐渐成为工业和学术领域的研究热点。为解决实际工业生产中由于产品缺陷数据集样本不足、类别不均衡导致的模型过拟合、检测精度低等问题，提出了一种基于Faster R-CNN算法框架优化的缺陷检测模型Faster R-CNN-H-BFC。该模型通过基于多层感知器（multi-layer perceptron, MLP）实现的幻觉网络，能够从具有丰富样本的基类中学习到类共享特征，并为新类生成额外的幻觉样本供模型训练。同时，针对Faster R-CNN本身存在的识别精度低以及检测效果差等问题，将原始的VGG16主干网络替换为具有残差结构的ResNet50，并引入了特征金字塔网络（Feature Pyramid Networks, FPN）实现多尺度特征融合，还添加了卷积块注意力模块（Convolutional Block Attention Module, CBAM）来增强模型的特征提取能力。实验和数据表明，改进后的缺陷检测模型在极少样本场景下具有较好的检测效果，平均检测精度相较于改进前提升了3.11%。
Abstract:
With the development of automatic detection technology, defect detection technology based on deep learning is gradually becoming a research hotspot in industrial and academic fields due to its high precision, high efficiency, and non-contact characteristics. In order to solve the problems of model overfitting and low detection accuracy caused by insufficient samples and unbalanced categories of product defect datasets in actual industrial production, a defect detection model Faster R-CNN-H-BFC based on the Faster R-CNN algorithm framework optimization is proposed. The illusion network based on multi-layer perceptron (MLP) can learn class sharing features from base classes with rich samples and generate additional illusion samples for new classes for model training. Aiming at the problems of low recognition accuracy and poor detection effect of Faster R-CNN, the original VGG16 backbone network is replaced with ResNet50 with residual structure, and Feature Pyramid Networks (FPN) is introduced to realize multi-scale feature fusion, and the Convolutional Block Attention Module (CBAM) is added to enhance the feature extraction ability of the model. Experiments and data show that the improved defect detection model has better detection effect in very few sample scenarios, and the average detection accuracy is 3.11% higher than that before improvement.

参考文献/References:

[1] 程锦锋, 方贵盛, 高惠芳. 表面缺陷检测的机器视觉技术研究进展[J]. 计算机应用研究, 2023, 40(4): 967-977.

[2] 杨传礼, 张修庆. 基于机器视觉和深度学习的材料缺陷检测应用综述[J]. 材料导报, 2022, 36(16): 226-234.
[3] LIFE-FEI, FERGUS, PERONA. A Bayesian approach to unsupervised one-shot learning of object categories[C]. Proceedings Ninth IEEE International Conference on Computer Vision. Nice, France: IEEE, 2003: 1134-1141.
[4] 史燕燕, 史殿习, 乔子腾, et al. 小样本目标检测研究综述[J]. 计算机学报, 2023, 46(8): 1753-1780.
[5] YANX, CHENZ, XUA, et al. MetaR-CNN: towards general solver for instance-level low-shot learning[C]. 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Seoul, Korea (South): IEEE, 2019: 9576-9585.
[6] JINGJ, MAH, ZHANGH. Automatic fabric defect detection using deep convolutional neural network[J]. Coloration Technology, 2019, 135(3): 213-223.
[7] VALENTEAC, WADAC, NEVESD, et al. Print defect mapping with semantic segmentation[C]. 2020 IEEE Winter Conference on Applications of Computer Vision (WACV). Snowmass Village, CO, USA: IEEE, 2020: 3540-3548.
[8] YANGY, WEIF, SHIM, et al. Restoring negative information in few-shot object detection[J]. 2020.
[9] WANGYX, GIRSHICKR, HEBERTM, et al. Low-shot learning from imaginary data[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT: IEEE, 2018: 7278-7286.
[10] VINYALSO, BLUNDELLC, LILLICRAPT. Matching networks for one-shot learning[C]. 30th Annual Conference on Neural Information Processing Systems, NIPS 2016: 3637-3645.
[11] SANTOROA, BARTUNOVS, BOTVINICKM, et al. Meta-learning with memory-augmented neural networks[C]. Proceedings of the 33rd International Conference on Machine Learning. PMLR, 2016: 1842-1850.
[12] WOOS, PARKJ, LEEJY, et al. CBAM: Convolutional block attention module[M]. FERRARI V, HEBERT M, SMINCHISESCU C, et al. Computer Vision–ECCV 2018: Cham: Springer International Publishing, 2018: 3-19.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期：2024-03 -17第一作者：何军红(1971 — )，男，浙江义乌人，博士，副教授，研究方向为工业互联网、智能制造、嵌入式系统开发。
更新日期/Last Update: 1900-01-01