基于改进ShuffleNetV2的男西服领型识别与分类模型
Recognition and classification of men's suit collar shape based on improved ShuffleNetV2
郑攀 1庹武 2魏新桥 2高雅昆 3杜聪 2刘思雨 2张欣汝2
作者信息
- 1. 河南科技学院 服装学院,河南 新乡 453003
- 2. 中原工学院 服装学院,河南 郑州 451191
- 3. 河南工学院 电气工程与自动化学院,河南 新乡 453003
- 折叠
摘要
针对男西服领型款式识别分类效果不佳的问题,提出一种基于改进ShuffleNetV2 的识别与分类模型.首先收集男西服样本图像,建立包含 9 类男西服领型的样本库;其次以ShuffleNetV2 模型为基础,引入 CBAM 注意力机制、H-Swish 激活函数来增强模型特征提取能力,有效防止神经元坏死;另外利用迁移学习的方法再次增强模型的特征提取能力,提高模型的识别适应性和识别精度;最后对样本库数据集进行数据增强后开始训练验证.实验结果对比表明,本文的改进模型可以实现对男西服领型的有效识别和分类,平均准确率达到 92.82%.本文为解决男西服领型识别与分类困难等问题提供了有效的解决方案,对服装定制行业的分类识别具有借鉴意义.
Abstract
To address the issue of poor recognition and classification of men's suit collar styles,a recognition and classification model based on an improved ShuffleNetV2 was proposed.Firstly,sample images of men's suits were collected to establish a sample library containing nine categories of men's suit collar styles.Secondly,the ShuffleNetV2 model was enhanced by introducing the CBAM attention mechanism and the H-Swish activation function to improve the model's feature extraction capability and effectively prevent neuron death.Additionally,transfer learning was utilized to further enhance the model's feature extraction ability,improving its recognition adaptability and accuracy.Finally,after performing data augmentation on the sample library dataset,training and validation were conducted.Comparative experimental results indicate that the improved model proposed in this paper can effectively recognize and classify men's suit collar styles,achieving an average accuracy rate of 92.82%.This paper provides an effective solution to the problem of recognizing and classifying men's suit collar styles,offering valuable insights for the custom clothing industry.
关键词
男西服领型/ShuffleNetV2/CBAM/H-Swish/迁移学习Key words
men's suit collar shape/ShuffleNetV2/CBAM/H-Swish/transfer learning引用本文复制引用
基金项目
河南省高等学校重点科研项目(19A540004)
河南省高等学校重点科研项目(23A540007)
教育部产学协同育人项目(220900693082010)
出版年
2024
NETL
NSTL
万方数据