遇见数据集

Digitized Metadata of Solar Disk H-alpha Observations At Sacramento Peak Observatory

收藏
Zenodo2026-04-20 更新2026-05-26 收录
官方服务:

资源简介:

The dataset contains metadata (date and time of observations) extracted from the digitized (scanned) photographic images of solar disk observations in neutral hydrogen H-alpha spectral line made at the U.S. National Solar Observatory at Sacramento Peak during 1966-2002 (https://nispdata.nso.edu/ftp/flare_patrol_h_alpha_sp/). Digitized text information is presented in a CSV table. The table includes the following columns: Filename: string with the original file name; Date: corrected date and time in the format YYYYMMDD_hhmmss; Day: day number predicted by the CRNN model; Month: month number inferred manually; Year: year number inferred manually if month is given and predicted by the CRNN model otherwise; Date_confidence: confidence in date digitization by the CRNN model; Date_box_confidence: confidence in date bounding box predicted by the YOLO model; Hour: hour number predicted by the CRNN model; Hour_confidence: confidence in hour digitization by the CRNN model; Hour_box_confidence: confidence in hour bounding box predicted by the YOLO model; Minutes: minutes number predicted by the CRNN model; Minutes_confidence: confidence in minutes digitization by the CRNN model; Minutes_box_confidence: confidence in minutes bounding box predicted by the YOLO model; Seconds: seconds number predicted by the CRNN model; Seconds_confidence: confidence in seconds digitization by the CRNN model; Seconds_box_confidence: confidence in seconds bounding box predicted by the YOLO model. Note that the table contains empty cells in cases where the YOLO model did not find the corresponding text field. Unknown dates and times are filled with 9999 for years and 99 for other values. Description of the text detection and recognition methods is given in the paper https://doi.org/10.1038/s41597-026-07243-x. Source code of the models is on https://github.com/observethesun/halpha.

创建时间:
2026-01-27
二维码
社区交流群
二维码
科研交流群
商业服务