收稿日期: 2022-01-22
网络出版日期: 2022-05-27
基金资助
国家重点研发计划资助项目(2018YFB0704400);国家重点研发计划资助项目(2020YFB0704503);上海市自然科学基金资助项目(20ZR1419000);之江实验室科研攻关资助项目(2021PE0AC02)
Anti-counterfeit label detection algorithm based on lightweight network
Received date: 2022-01-22
Online published: 2022-05-27
近年来, 伪造盗版产品带来的经济损失逐年增大, 伪造技术不断提升, 防伪检测问题受到了广泛关注. 为了解决现有防伪检测方法的计算量大、资源占用高、检测耗时较长等问题, 提出了一种基于轻量级网络的防伪标签识别检测模型, 该模型采用更为轻量的卷积神经网络(convolutional neural network, CNN)来进行形状和纹理的识别. 在形状识别任务中, 降低池化层大小以增强模型学习能力; 在纹理分类任务中, 使用协调注意力(coordinate attention, CA)模块来增强模型对单一特征图的信息获取. 通过设计损失函数增强模型对真伪样本识别能力, 并通过特征向量最大值得到预测结果. 实验结果表明, 该方法整体识别检测的准确率可达 95.67%, 检测时间相较于传统方法有显著减少.
张宏坤, 韩越兴, 陈侨川, 巫金波 . 基于轻量级网络的防伪标签检测算法[J]. 上海大学学报(自然科学版), 2022 , 28(3) : 534 -544 . DOI: 10.12066/j.issn.1007-2861.2373
Economic loss caused by counterfeit and pirated products has increased annually; in this regard, counterfeiting technology has been improved continuously, where researchers attempt to improve anti-counterfeiting detection. To alleviate the complex computation, high resource requirement, and long detection time of existing anti-counterfeiting detection methods, an anti-counterfeiting label identification and detection model based on a lightweight network is proposed herein. Convolutional neural networks (CNNs) are adopted for shape and texture recognition. During shape recognition, the size of the pooling layer is reduced to enhance model learning ability. A coordinate attention (CA) module is used in texture classification to enhance the information acquisition of a single feature graph. The loss function is designed to enhance the model's ability to identify authentic samples. Finally, the prediction result is obtained using the maximum feature vector. Experimental results show that the maximum overall detection accuracy achievable by the proposed method is 95.67%, and that the detection time improved significantly as compared with the conventional method.
| [1] | Carro-Temboury M R, Arppe R, Vosch T, et al. An optical authentication system based on imaging of excitation-selected lanthanide luminescence[J]. Science Advances, 2018, 4(1): e1701384. |
| [2] | Hu Z Y, Comeras J M M L, Park H, et al. Physically unclonable cryptographic primitives using self-assembled carbon nanotubes[J]. Nature Nanotechnology, 2016, 11(6): 559-565. |
| [3] | Economics F. The economic impacts of counterfeiting and piracy[EB/OL]. [2021-12-31]. http://zhuanlan.zhihu.com/p/25110213. |
| [4] | Tang G, Chen L, Wang Z, et al. Faithful fabrication of biocompatible multicompartmental memomicrospheres for digitally color-tunable barcoding[J]. Small, 2020, 16(24): 1907586. |
| [5] | Kim M S, Lee G J, Leem J W, et al. Revisiting silk: a lens-free optical physical unclonable function[J]. Nature Communications, 2022, 13(1): 1-12. |
| [6] | Liu Y, Han F S, Li F S, et al. Inkjet-printed unclonable quantum dot fluorescent anti-counterfeiting labels with artificial intelligence authentication[J]. Nature Communications, 2019, 10(1): 1-9. |
| [7] | Lin Y H, Zhang H K, Feng J Y, et al. Unclonable micro-texture with clonable micro-shape towards rapid, convenient, and low-cost fluorescent anti-counterfeiting labels[J]. Small, 2021, 17(30): 2100244. |
| [8] | Han F, Liu Y, Li F S, et al. Self-assembly of coordination polymers on plasmonic surfaces for computer vision decodable, unclonable and colorful security labels[J]. Journal of Materials Chemistry C, 2019, 7(42): 13040-13046. |
| [9] | Gu Y Q, He C, Zhang Y Q, et al. Gap-enhanced Raman tags for physically unclonable anticounterfeiting labels[J]. Nature Communications, 2020, 11(1): 1-13. |
| [10] | Bae H J, Bae S, Park C, et al. Biomimetic microfingerprints for anti-counterfeiting strategies[J]. Advanced Materials, 2015, 27(12): 2083-2089. |
| [11] | Zheng Y H, Cheng J, Ng S H, et al. Unclonable plasmonic security labels achieved by shadow-mask-lithography-assisted self-assembly[J]. Advanced Materials, 2016, 28(12): 2330-2336. |
| [12] | Jing L, Xie Q, Li H L, et al. Multigenerational crumpling of 2D materials for anticounterfeiting patterns with deep learning authentication[J]. Matter, 2020, 3(6): 2160-2180. |
| [13] | Zheng X, Zhu Y B, Liu Y, et al. Inkjet-printed quantum dot fluorescent security labels with triple-level optical encryption[J]. ACS Applied Materials & Interfaces, 2021, 13(13): 15701-15708. |
| [14] | 葛道辉, 李洪升, 张亮, 等. 轻量级神经网络架构综述[J]. 软件学报, 2020, 31(9): 2627-2653. |
| [15] | Ma N, Zhang X, Zheng H T, et al. Shufflenet v2: Practical guidelines for efficient cnn architecture design[C]// Proceedings of the European Conference on Computer Vision (ECCV). 2018: 116-131. |
| [16] | Tan M X, Le Q V. Efficientnet: rethinking model scaling for convolutional neural networks[C]// International Conference on Machine Learning. 2019: 6105-6114. |
| [17] | Howard A, Sandler M, Chu G, et al. Searching for mobilenetv3 [C]// roceedings of the IEEE/CVF International Conference on Computer Vision. 2019: 1314-1324. |
| [18] | Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2018: 7132-7141. |
| [19] | Hou Q B, Zhou D Q, Feng J S. Coordinate attention for efficient mobile network design[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021: 13713-13722. |
| [20] | Chattopadhay A, Sarkar A, Howlader P, et al. Grad-cam++: generalized gradient-based visual explanations for deep convolutional networks[C]// 2018 IEEE Winter Conference on Applications of Computer Vision (WACV). 2018: 839-847. |
/
| 〈 |
|
〉 |
