<p>Breast cancer is a lethal illness and the most frequent disease impacting women all over the world. Appropriate classification and the earliest classification of this severe disease are important. The accurate and timely recognition of breast cancer with Histopathological Images may eradicate the mortality rate and secure humans from more damage. Former techniques employed for breast cancer classification are highly expensive on the basis of cost as well as time. To suppress these issues, histological structure-based breast cancer classification using Histopathological images is established as Mobile Siamese Forward Harmonic Net (MSFHNet). The denoising process is performed by Medav filter and then for histological structure segmentation, a Pyramidal Attention-based Deep Learning network (Lung_PAYNet) is used. After the segmentation, feature extraction is conducted. Moreover, the histological structure of breast cancer is classified as mitosis, apoptosis, tumor nuclei, non-tumor nuclei, tubule, and non-tubule, where this classification is accomplished by MSFHNet. Finally, the localization of histological structure is conducted. While evaluating the model with some evaluation metrics, it obtained an Accuracy of 98.724%, Sensitivity of 97.586%, and Specificity of 98.417%.</p>