To address the lack of attention to frequency features in existing image classification networks, we propose an image classification network with frequency modulation mechanism (FMM-Net). First, we develop Context-Adaptive Aware Convolution (CAAC) to adaptively capture information such as object edges and details. Second, we introduce Frequency-Modulated Convolution (FMC), which processes frequency information through an enable-learning scaling factor to adjust the contribution of input frequency information, thereby enabling effective extraction of image frequency features. Next, we design Frequency-Modulated Attention (FMATT), which enables the model to dynamically focus on and activate frequency features while extracting spatial features. Finally, based on the difference-aware module, Frequency-Modulated Convolution, and Frequency-Modulated Attention, we construct an image classification network with a frequency modulation mechanism. Compared to current state-of-the-art models, the proposed network significantly improves image classification performance.