Cancer is the leading cause of cancer-related deaths universal. When it comes to annual killers, breast cancer is right up there. Current approaches employ Inception as a mobile net classification module; the suggested technique is the m-Xception-residual exemplary. In contrast, the suggested model extracts features for classification using a SoftMax layer that is based on logarithms. The accuracy numbers of the suggested classical reflect the fact that these improvements substantially decrease false-positive answers, which in turn boost real positives. The parameters are ideally modified by the ISOA model to fine-tune the suggested model. In this article, we show you how to pick the right features for the HER2 score diagnostic problem so you can perform better. Using HER2 images—which are distinct in structure and morphology—we demonstrated that picking the right characteristics might enhance the accuracy of the model's presentation in breast cancer imaging. Furthermore, the temporal complexity of the picture assessment was lowered by lowering the amount of characteristics. One possible application of the planned study is the remote diagnosis of breast cancer in patients. The proposed effort could be much better if it makes use of other famous DL and TL ideas.