计算机科学
图像(数学)
人工智能
大数据
模式识别(心理学)
算法
数据挖掘
作者
Minghui Jin,Zihui Dou,Haobin Xu,Feng Lu
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2025-05-23
卷期号:: 1186-1194
标识
DOI:10.1109/icetci64844.2025.11084128
摘要
This paper addresses the limitations of traditional ResNet in handling multi-scale features and complex image scenarios, which hinder its performance in image classification. To overcome these challenges, we propose CFResNet, an improved model incorporating batch normalization, feature pyramid networks, and channel attention mechanisms. CFResNet significantly enhances classification accuracy and robustness, particularly in complex scenarios and multi-scale feature fusion. Experimental results demonstrate its superiority, achieving 99.53% accuracy on a bird recognition dataset, outperforming traditional ResNet and other improved models. This research offers an efficient and innovative solution for image classification, with broad application potential.
科研通智能强力驱动
Strongly Powered by AbleSci AI