This dataset is an addition to the Cholec80 cholecystectomy dataset (https://doi.org/10.1109/TMI.2016.2593957). The archive contains mapping files for specific video segments containing smoke in various intensities and they are split up into two classes (see labels.csv): no smoke no visual signs of smoke smoke smoke in various intensities The dataset altogether comprises 101054 images (50501 no smoke + 50553 smoke) and the archive's subfolder cholec80_mappings is organized as follows: labels.csv mapping: label number -> class index.csv index linking source videos of Cholec80 with corresponding mapping files. Contains video path (relative to Cholec80 archive), fps, mapping file path (relative to index.csv) and frame counts per video for both labels. videoXX.mp4.csv frame mapping for video XX: start frame, end frame, label You are kindly requested to cite the original work that led to the creation of the dataset: https://doi.org/10.1145/3126686.3126690. Additionally, please adhere to the usage rights of the base dataset that was used to create these mappings, i.e. Cholec80 (http://camma.u-strasbg.fr/datasets). For the latest updates, please visit the dataset's homepage: http://ftp.itec.aau.at/datasets/Smoke_cholec80/.

{"references": ["Leibetseder et al.(2017), Real-Time Image-based Smoke Detection in Endoscopic Videos, https://doi.org/10.1145/3126686.3126690"]}