ResNet-SE-CBAM 罗网络为少数射击和不平衡的PCB缺陷分类
Chao-Hsiang Hsiao1, Huan-Che Su2, Yin-Tien Wang3,4
1Department of Computer Science and Information Engineering, Tamkang University, New Taipei City 251301, Taiwan.
Sensors (Basel, Switzerland)
|July 12, 2025
概括
本研究引入了一种使用ResNet-SE-CBAM语网络进行产品缺陷检测的新型几次学习方法. 该方法提高了准确性,降低了错误率,即使数据有限,这使得它非常适合工业应用.
科学领域:
- 计算机科学 计算机科学
- 人工智能的人工智能
- 机器学习 机器学习
背景情况:
- 大规模生产缺陷检测面临着小,不平衡的数据集的挑战,限制了传统的深度学习.
- 短暂的学习对于将模型适应实际的工业场景而使用最小的数据至关重要.
研究的目的:
- 用有限的数据开发和评估一款用于有效检测产品缺陷的几次性学习模型.
- 提高模型的概括性,稳定性和在工业环境中的应用性.
主要方法:
- 提出了一个ResNet-SE-CBAM语网络用于特征提取,结合注意力机制和度量学习.
- 用于嵌入学习的三重损失和用于样本选择的结构相似度指数 (SSIM) 测量.
- 实施了高缺陷率的培训策略和K-Nearest Neighbor (KNN) 分类器,以提高稳定性和减少假阴性.
主要成果:
- 实现了94%的分类准确度和2%的假负率 (FNR),良好的缺陷比率为20:40.
- 当缺陷样本数增加到80时,达到零假阴性 (FNR = 0%).
- 在准确性和错误率方面超过了传统的深度学习模型,如YOLO.
结论:
- 与传统的深度学习模型相比,拟议的度量学习方法在短时间内检测缺陷方面表现出卓越的性能.
- 该系统提供了高可靠性和工业部署的潜力,有效地应对有限和不平衡数据集的挑战.
相关概念视频
01:26Classification of Systems-I
314
Linearity is a system property characterized by a direct input-output relationship, combining homogeneity and additivity.
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:
Homogeneity dictates that if an input x(t) is multiplied by a constant c, the output y(t) is multiplied by the same constant. Mathematically, this is expressed as:
314
01:31Classification of Systems-II
242
Continuous-time systems have continuous input and output signals, with time measured continuously. These systems are generally defined by differential or algebraic equations. For instance, in an RC circuit, the relationship between input and output voltage is expressed through a differential equation derived from Ohm's law and the capacitor relation,
242
01:30Classification of Signals
903
In signal processing, signals are classified based on various characteristics: continuous-time versus discrete-time, periodic versus aperiodic, analog versus digital, and causal versus noncausal. Each category highlights distinct properties crucial for understanding and manipulating signals.
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
A continuous-time signal holds a value at every instant in time, representing information seamlessly. In contrast, a discrete-time signal holds values only at specific moments, often denoted as x(n), where...
903