Polyps on the colon’s surface are the starting point for colon cancer, a common and potentially fatal type of cancer. Early identification of these polyps and cancers is essential for efficient screening. Deep Convolutional Neural Network (DCNN) methodologies, particularly utilizing U-Net architecture, are implemented to enhance colorectal cancer detection. Data augmentation enhances the dataset and enables accurate detection. Current polyp segmentation approaches are challenged by irregular color patterns, inconsistent polyp presence, and reflection effects in colonoscopy and endoscopy pictures, which may result in either over- or under-segmentation. In order to improve polyp detection, this work suggests an enhanced U-Net-based segmentation algorithm that incorporates edge and valley detection techniques. To increase the accuracy of detection, post-processing algorithms combine together neighboring items and enhance segmentation boundaries. Doctors can diagnose polyps more quickly, accurately, and easily using this computerized procedure. The development of Computer-Aided Diagnosis (CAD) systems for automated polyp diagnosis depends on these techniques. In this context, our suggested U-Net approach provides an efficient DCNN design solution, improving automated polyp detection techniques.