现代计算机2024,Vol.30Issue(1) :75-78.DOI:10.3969/j.issn.1007-1423.2024.01.013

基于Xception和迁移学习的图像分类研究

Research on image classification based on Xception and transfer learning

谢生锋
现代计算机2024,Vol.30Issue(1) :75-78.DOI:10.3969/j.issn.1007-1423.2024.01.013

基于Xception和迁移学习的图像分类研究

Research on image classification based on Xception and transfer learning

谢生锋1
扫码查看

作者信息

  • 1. 河南工学院计算机科学与技术学院,新乡 453003
  • 折叠

摘要

卷积神经网络CNN在图像处理中应用非常广泛,通过CNN在ImageNet数据集上训练出了AlexNet、VGGNet、ResNet和Xception等经典深度学习模型.通过迁移学习将Xception模型作为预训练模型,使用Xception模型的卷积基对Kaggle平台上的猫狗数据集进行特征提取,并对Xception模型进行微调,采用TensorFlow框架实现了猫狗图像的准确识别.

Abstract

Convolutional neural network(CNN)is widely used in image processing,and classical deep learning models such as AlexNet,VGGNet,ResNet and Xception are trained by CNN on ImageNet dataset.Through transfer learning,the Xception model is used as a pre-trained model,and the convolution basis of the Xception model is used to extract features from the cat and dog data set on Kaggle platform,and TensorFlow framework is used to achieve accurate identification of cat and dog images.

关键词

CNN/Xception/迁移学习/图像分类/TensorFlow

Key words

CNN/Xception/transfer learning/image classification/TensorFlow

引用本文复制引用

出版年

2024
现代计算机
中大控股

现代计算机

影响因子:0.292
ISSN:1007-1423
参考文献量5
段落导航相关论文