Supplementary Materials for REAL-Colon: A Dataset for Developing Real-World AI Applications in Colonoscopy

Supplementary Material for our submission: REAL-Colon: A dataset for developing real-world AI applications in colonoscopy.

Access the dataset here: https://figshare.com/articles/media/REAL-colon_dataset/22202866

Video 003-013.mp4: shows a 60-minute colonoscopy video featuring 6 polyps, the longest in the paper test set, demostrating the detection capabilities of the best off-the-shelf SSD model trained in our submission (with predictions marked in cyan and ground truth boxes in white). The input resolution was reduced to a 300x300 image size, but the original frame rate was maintained.

Polyp_Annotation_Tool.mp4: This brief video demonstrates the bounding box annotation process using COSMO IMD’s proprietary annotation tool, which was utilized to annotate the entire dataset.

Paper Pre-Print:
Biffi, C., Antonelli, G., Bernhofer, S., Hassan, C., Hirata, D., Iwatate, M., Maieron, A., Salvagnini, P., and Cherubini, A., 2024. REAL-Colon: A Dataset for Developing Real-World AI Applications in Colonoscopy. arXiv preprint arXiv:2403.02163.