KvasirSEG

Kvasir-SEG is an open-access dataset of 1000 gastrointestinal polyp images from Kvasir Dataset v2 with pixel-wise polyp segmentation masks, manually annotated and verified by an experienced gastroenterologist (Simula Research Laboratory; Jha et al., MMM 2020). Image resolution varies from 332x487 to 1920x1072 pixels; the upstream release stores JPEG images and masks under the same filenames in two folders, plus polyp bounding boxes in a JSON file. This mirror serves the 800/100/100 ESFPNet cut with the masks binarised to ids 0/1.

Classes

id	name
0	Background
1	Polyp

License

Kvasir-SEG Terms of Use (research and educational use; commercial use needs written permission; citation required) (https://datasets.simula.no/kvasir-seg/)

Terms of use

Quoted verbatim from the official release page (https://datasets.simula.no/kvasir-seg/, section "Terms of use"; the authors' GitHub README, https://github.com/DebeshJha/Kvasir-SEG, carries the same terms in near-identical wording):

The use of the Kvasir-SEG dataset is restricted for research and educational purposes. The use of the Kvasir-SEG dataset for commercial purposes is forbidden without prior written permission. For other purposes, contact us (see below). In all documents and publications that use the Kvasir-SEG dataset or report experimental results based on the Kvasir-SEG dataset, a reference to the dataset paper has to be included (see below).

This mirror is served for non-commercial research and educational use under those terms (MedOtter card redistribution.allowed: yes_noncommercial); commercial use requires the authors' prior written permission, and every publication using the data must cite the dataset paper below.

Citation

Jha, D., Smedsrud, P. H., Riegler, M. A., Halvorsen, P., de Lange, T., Johansen, D., & Johansen, H. D. (2020). Kvasir-SEG: A Segmented Polyp Dataset. In MultiMedia Modeling (MMM 2020), Lecture Notes in Computer Science, vol. 11962, pp. 451-462. Springer. https://doi.org/10.1007/978-3-030-37734-2_37