Research on Big Data Image Classification Based on CFResNet Algorithm

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.