对于金属表面缺陷的短拍跨插件自适应性内存语义细分部分化
Jiyan Zhang1, Hanze Ding1, Ming Peng1
1College of Mathematics and Information Engineering, Longyan University, Longyan, 364012, China.
这项研究引入了用于金属表面缺陷检测的情节适应性记忆网络 (EAMNet). EAMNet通过适应微妙的训练变化,提高缺陷区域分析和细分精度来改进少数镜头的语义细分.
科学领域:
- 计算机视觉 计算机视觉
- 机器学习 机器学习
- 材料科学 材料科学 材料科学
背景情况:
- 短暂的语义细分对于具有有限数据的金属表面缺陷检测至关重要.
- 现有的方法在少数场景中扎着适应性和细分细分度.
研究的目的:
- 提出一个情节适应性记忆网络 (EAMNet),以改善金属表面缺陷检测中的几次射击语义细分.
- 解决以前方法在语义描述和细分细分度方面的局限性.
主要方法:
- 开发了一个使用适应因子进行跨插件语义依赖模型的插件适应性记忆单元 (EAMU).
- 引入了一个上下文适应模块 (CAM) 通过聚合层次特征进行细粒度细分.
- 拟议的全球响应面膜平均聚合 (GRMAP) 用于直接细粒度线索提取.
- 实施注意力蒸 (AD) 来稳定跨情节的适应.
主要成果:
- 对于金属表面缺陷,EAMNet在短时间的语义细分方面表现出卓越的性能.
- 提出的方法在表面缺陷和FSSD-12数据集上取得了最先进的结果.
- 观察到适应性和细分细粒度的显著改善.
结论:
- 在短暂学习中,EAMNet有效地处理训练情节之间的微妙变化.
- 拟议的方法提高了金属表面缺陷细分的准确性和颗粒度.
- 这项工作为工业缺陷检测中的短拍语义细分设定了新的基准.
更多相关视频
07:59Using Practice Testing, Public Speaking, and Source Monitoring to Examine the Influences of Learning Strategies and Stress on Episodic Memory
Published on: June 14, 2019
08:17A Semantic Priming Event-related Potential ERP Task to Study Lexico-semantic and Visuo-semantic Processing in Autism Spectrum Disorder
Published on: April 12, 2018
相关概念视频
08:16Combining Behavior and EEG to Study the Effects of Mindfulness Meditation on Episodic Memory
07:59Using Practice Testing, Public Speaking, and Source Monitoring to Examine the Influences of Learning Strategies and Stress on Episodic Memory
08:17A Semantic Priming Event-related Potential (ERP) Task to Study Lexico-semantic and Visuo-semantic Processing in Autism Spectrum Disorder
09:13A Real-world What-Where-When Memory Test
Language: The N400 in Semantic Incongruity
Understanding language is one of the most complex cognitive tasks that humans are capable of. Given the incredible amount of possible choices when combining individual words to form meaning in sentences, it is crucial that the brain is able to identify when words form coherent combinations and when an anomaly appears that undermines meaning. Extensive research has shown that certain...
11:01Examining the Characteristics of Episodic Memory using Event-related Potentials in Patients with Alzheimer's Disease
