Image classification has witnessed remarkable progress due to the advent of Convolutional Neural Networks [1] (CNNs). However, the capacity to focus on salient information within images remains a challenge. This research explores the integration of Channel Attention and Spatial Attention mechanisms into baseline CNN architectures to enhance classification performance. We conduct experiments on standard image classification datasets to evaluate the efficacy of integrating CBAM [13] (Convolutional Block Attention Module), BAM [11] (Bottleneck Attention Module), and scSE [12] (spatial and channel Squeeze & Excitation) against baseline models. Experimental results demonstrate that leveraging these attention modules significantly improves classification accuracy.