To construct a system for autonomously recognising cancer, this paper uses an integration technique that includes CNN and picture texture attribute extraction. A customised Deep Dense Convolution Neural Network is used to analyse an input image in the CNN stage. To improve the efficacy of classification in the extraction-based phase, texture features are established using the Curve let Transform. Internally, the built CNN Deep dense classifier collects features from the enhanced image and classifies them as regular or irregular tumour imaging. Furthermore, the proposed Deep net classifier for CNN analyses the observed image of tumour as (low intensity) benign (high grade) malignant based on its component properties. The recommended CNN's strong performance suggests that it might be utilised as an automated approach for detecting tumours, potentially saving time and money. Reducing pathologists' workload. The proposed CNN classification approach includes variables such as sensitivity, specificity, and accuracy in order to detect and classify breast cancer effectively. According to the testing results, the average sensitivity is 94.5 percent, the average specificity is 98.16 percent, the average accuracy is 97.25 percent, and the FAR is 1.83 percent.