Anomaly Detection of TP4056 Charging Module Based on Improved SimpleNet

DOI: https://doi.org/10.62517/jbdc.202501418

Author(s)

Shu Chen, Rouyi Fan, Xiaofeng Li*

Affiliation(s)

School of Artificial Intelligence and Big Data, Henan University of Technology, Zhengzhou, Henan, China *Corresponding Author

Abstract

This study suggests an industrial anomaly detection technique based on an enhanced SimpleNet to fulfill the surface defect detection requirements of the TP4056 charging module in the realm of industrial manufacturing. First, the algorithm adds a Multi-Head Attention (MHA) module to SimpleNet’s discriminator. By concentrating on important details within picture features, this module can improve the model’s capacity to distinguish intricate features and precisely identify a range of minute flaws in the TP4056 charging module. Additionally, residual connections are used to optimize the gradient propagation path. This preserves original information while injecting attention-guided feature interactions by adding the original features to the attention correction term. A dropout module, which successfully avoids overfitting and enhances the model’s generalization ability, is presented in consideration of the model’s capacity for generalization. In order to optimize the anomalous feature production process and enable the model to learn anomaly feature representations under various noise situations, SimpleNet’s anomaly feature generator simultaneously implements a dynamic noise correction technique. The VISA-PCB4 dataset was used to verify the model’s efficacy. According to experiments, the enhanced model performs better on image-level AUROC, pixel-level AUROC, and anomaly pixel-level AUROC (PRO) measures, achieving 95.2%, 97.3%, and 88.8%, respectively. When compared to the original model, the anomaly pixel-level AUROC rose by 9.8%, greatly improving the model’s capacity to identify anomalous pixels and offering a practical way to check the quality of charging modules.

Keywords

Industrial Quality Inspection; SimpleNet; Attention Mechanism; Dynamic Noise

References

[1] Chandola V, Banerjee A, Kumar V. Anomaly detection: A survey. ACM computing surveys (CSUR), 2009, 41(3): 1-58. [2] Liu J, Xie G, Wang J, et al. Deep industrial image anomaly detection: A survey. Machine Intelligence Research, 2024, 21(1): 104-135. [3] Wang Baoyu. Research on Solder Joint Defect and Component Detection Method Based on Deep Learning. Nanchang University, 2020. [4] Zhong Zhuoran. Research and Application of Artificial Intelligence Image Processing Technology Based on Deep Learning. Internet Weekly, 2025, (14): 50-52. [5] Guo Jialin, Zhi Min, Yin Yanjun, et al. A Survey of Hybrid Models of CNN and Visual Transformer in Image Processing. Journal of Computer Science and Exploration, 2025, 19(01): 30-44. [6] Kim Y. Convolutional Neural Networks for Sentence Classification. Eprint Arxiv, 2014. [7] Roth K, Pemula L, Zepeda J, et al. Towards Total Recall in Industrial Anomaly Detection. 2021. DOI:10.48550/arXiv.2106.08265. [8] Defard T, Setkov A, Loesch A, et al. PaDiM: A Patch Distribution Modeling Framework for Anomaly Detection and Localization. 2021. [9] Liu Z, Zhou Y, Xu Y, et al. Simplenet: A simple network for image anomaly detection and localization. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2023: 20402-20411. [10]Gong D, Liu L, Le V, et al. Memorizing normality to detect anomaly: Memory-augmented deep autoencoder for unsupervised anomaly detection. Proceedings of the IEEE/CVF international conference on computer vision. 2019: 1705-1714. [11]Ristea N C, Madan N, Ionescu R T, et al. Self-supervised predictive convolutional attentive block for anomaly detection. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2022: 13576-13586. [12]Zavrtanik V, Kristan M, Skočaj D. Reconstruction by inpainting for visual anomaly detection. Pattern Recognition, 2021, 112: 107706. [13]Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection. Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2117-2125. [14]Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need. Advances in neural information processing systems. 2017, 30. [15]He K, Zhang X, Ren S, et al. Deep residual learning for image recognition. Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778. [16]Srivastava N, Hinton G, Krizhevsky A, et al. Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research, 2014, 15(1): 1929-1958.