Kvasir-SEG is an open-access dataset of gastrointestinal polyp images and corresponding segmentation masks, manually annotated by a medical doctor and verified by an experienced gastroenterologist. The dataset also includes bounding boxes for the polyp regions. The authors demonstrate the use of the dataset with both traditional segmentation approaches and a deep-learning-based Convolutional Neural Network (CNN) approach, specifically Fuzzy C-mean clustering (FCM) and Deep Residual U-Net (ResUNet). The dataset is valuable for researchers to reproduce results and compare methods, particularly in the field of polyp segmentation and automatic analysis of colonoscopy images. The paper discusses the motivation behind the dataset, the evaluation metrics used, and the experimental results, showing that the ResUNet model outperforms the FCM clustering in terms of Dice coefficient and mean Intersection over Union (IoU). The Kvasir-SEG dataset is released to the multimedia and medical research communities to facilitate further development and evaluation of computer vision methods.Kvasir-SEG is an open-access dataset of gastrointestinal polyp images and corresponding segmentation masks, manually annotated by a medical doctor and verified by an experienced gastroenterologist. The dataset also includes bounding boxes for the polyp regions. The authors demonstrate the use of the dataset with both traditional segmentation approaches and a deep-learning-based Convolutional Neural Network (CNN) approach, specifically Fuzzy C-mean clustering (FCM) and Deep Residual U-Net (ResUNet). The dataset is valuable for researchers to reproduce results and compare methods, particularly in the field of polyp segmentation and automatic analysis of colonoscopy images. The paper discusses the motivation behind the dataset, the evaluation metrics used, and the experimental results, showing that the ResUNet model outperforms the FCM clustering in terms of Dice coefficient and mean Intersection over Union (IoU). The Kvasir-SEG dataset is released to the multimedia and medical research communities to facilitate further development and evaluation of computer vision methods.