Across the world, many people die from breast cancer, so spotting it early helps keep survival chances high. We present in this paper a deep learning application using the HSEffNet7 (Hybrid SE-ResNet with EfficientNet-B7) model for analyzing mammogram images. To classify benign and malignant tumors, the model depends on HSEffNet7, a feature-rich model shown to have success with big image datasets. The model’s success is clear by its accuracy which is 89.01%, precision of 80.19%, recall of 81.41% and an F1 score of 80.79%. A Dice coefficient of 0.8079 also indicates the credibility of the model in classifying malignant cases with high accuracy. The model accurately identified 5 out of 6 images of breast cancer while testing, proving the feasibility of the model in practical clinical applications. In spite of its encouraging results, one misclassification indicated the difficulties associated with handling dataset imbalance and mammogram image complexities. The proposed HSEffNet7 based model is a noteworthy advancement in automated breast cancer classification, demonstrating its potential in helping medical practitioners in early detection. Future research needs to emphasize handling issues such as dataset imbalance and enhancing model generalization in order to maximize effectiveness for practical applications.