Steganography is a security mechanism used to conceal sensitive data in various forms of digital media. This is useful for transmitting sensitive information without disclosing its existence. Our paper introduces an advanced steganographic method based on deep learning, which combines Convolutional Neural Networks (CNNs) with Quantization and Convolutional Block Attention Module (CBAM). Our steganography method addresses several issues with traditional approaches, such as low embedding capacity and steganographic image distortions. The integration of CNNs in our proposed approach allows for effective feature extraction, while Quantization refines and enhances the extracted features which helps to preserve the image quality after embedding. In addition, the CBAM attention mechanism captures useful features while filtering out noise and irrelevant information. This helps us to reduce the feature distortions and improve the feature representations. We have implemented and tested the effectiveness of our steganographic approach with various evaluation metrics. The experimental results show that our model outperforms the state-of-the-art approaches.