ABSTRACT
Gastroenterologists conduct examinations of abnormal tissue in the gastrointestinal tract, during which they perceive polysegmentation as a complex undertaking. While it provides valuable assistance to gastroenterologists. Polyps, which are abnormal growths of tissue, fundamentally create within the colorectal locale of the gastrointestinal tract. Moreover, it is pre-dominantly found within the mucous layer, which as of now has various bulges of the smaller scale abnormal tissue, altogether opening up the hazard of sickness advancement. Early location of polyps makes a difference repress their movement into cancerous tissue, such as adenomas, which may inevitably create into cancer. The anomalies watched incorporate colored lifted polyps, ordinary cecum, typical pylorus, ordinary z-line, esophagitis, polyps, ulcerative colitis, and colored resection edges. These anomalies are detected using the pre-processed encoder and decoder stages. The quality of colonoscopy images can be significantly enhanced with the implementation of the preprocessing phase. The gastrointestinal tract primarily exhibits abnormalities such as polyps and colitis. The VGG convolutional model provides the highest level of accuracy compared to other current deep learning models. For this research article, we utilized the Kvasir data collection. The Kvasir dataset comprises 4,000 annotated photos of the gastrointestinal tract, categorized into 8 distinct classes. Each class has approximately 500 images. Data sets consist primarily of information regarding the bowel segment of the gastrointestinal tract during colonoscopy. The VGG model achieves superior performance on the Kvasir dataset, with accuracy ranging from a minimum of 87% to a maximum of 97%.
