As one of the most common cancers in women globally, breast cancer must be detected early to effectively treat the condition and provide better patient outcomes. Histopathology images, which provide detailed information about tissue morphology, are routinely used for cancer diagnosis. Deep learning methods have demonstrated impressive results in a range of medical image processing applications, such as cancer detection, in recent years. This research provides a novel approach that uses deep learning approaches to diagnose breast cancer from histopathology images. A pre-trained Convolutional Neutral Network (CNN) model is employed to determine if a pre-segmented breast cancer mass mammography tumor is benign or malignant. Based on a modified ensembled version of ResNet and DenseNet, the proposed system addresses the classification problems. A thorough testing and training process has been applied to the recommended architecture. CNN was trained using the RGB model dataset, which includes 2480 benign and 5429 malignant cases. The model acquires a 98 percent accuracy, 99 percent precision, and 97 percent recall rate for benign instances. For malignant patients, a 97 percent accuracy, 95 percent precision, and 96 percent recall rate are achieved.