AI training data set for Moon crater annotation with Crowd Science via a videogame

Overview This is the additional data for the paper 'Generating an AI training data set for Moon crater annotation with Crowd Science via a videogame' by Stephan Jacob* ([e-mail]), Paul Nasdalack ([e-mail]) and Marius Schwinning * main corresponding author Files All files contain data from the October 12th 2025. New data may be available, please contact the authors if you need further information 1_250912_craters_merged.csv This file contains the processed and merged craters with information about the NAC Tile they have marked and pixel offsets to the specific tile and within the tile. To get the final coordinate within the NAC Image, add the corresponding x and y values up to get an absolute coordinate. Additionally the file contains a confidence score based on how many users have contributed to a crater and how well they were marking. This can serve as a threshold value to filter out low confidence craters. 2_250912_craters_usersubmitted.csv This file contains the unprocessed user submitted craters with information about the NAC Tile they have marked and pixel offsets to the specific tile and within the tile. To get the final coordinate within the NAC Image, add the corresponding x and y values up to get an absolute coordinate 3_250912_users.csv This file contains information on the users, how many craters they have submitted and their final accuracy score概述 本数据集为Stephan Jacob*（主要通讯作者，邮箱：[e-mail]）、Paul Nasdalack（邮箱：[e-mail]）与Marius Schwinning合著论文《通过视频游戏结合众包科学生成月球陨石坑标注AI训练数据集》的补充数据。 文件说明 所有文件的数据均采集自2025年10月12日。若需获取更新数据，请联系论文作者获取进一步信息。 1. 1_250912_craters_merged.csv 该文件存储经处理与合并后的陨石坑数据，包含其标注所属的NAC瓦片（NAC Tile）信息，以及相对于该瓦片的像素偏移量。若需获取陨石坑在NAC影像中的绝对坐标，只需将对应x、y方向的偏移量相加即可得到。此外，文件中还包含一项置信度评分，该评分基于参与该陨石坑标注的用户数量及其标注质量计算得出，可作为筛选低置信度陨石坑的阈值参数。 2. 2_250912_craters_usersubmitted.csv 该文件存储未经处理的用户提交陨石坑数据，包含其标注所属的NAC瓦片信息，以及相对于该瓦片的像素偏移量。若需获取陨石坑在NAC影像中的绝对坐标，只需将对应x、y方向的偏移量相加即可得到。 3. 3_250912_users.csv 该文件存储用户相关信息，包括其提交的陨石坑总数及最终准确率评分。