Colorectal cancer (CRC) is one of the most common cancers with a significant mortality rate. Colonoscopy is the primary colorectal cancer screening method since it reduces CRC mortality. Considering this, a dependable computer-assisted polyp identification and classification system has the potential to considerably increase colonoscopy efficiency. Automated diagnosis utilizes computer-aided ways to analyze all the results quickly and correctly. In this paper, we used the Kvasir-SEG dataset to classify gastrointestinal disorder. The Kvasir dataset contains 5000 images divided evenly into five gastrointestinal tract-related groups: normal cecum, polyps, ulcerative colitis, dye-lifted polyps, and colored resection margins. By updating Efficient Model B0 and applying it to B7, we achieved 97% testing accuracy.