Help LabelMe: A Fast Auxiliary Method for Labeling Image and Using It in ChangE's CCD Data

Automatic detection of regions of interest in lunar image data is a basic subject for understanding the moon. In recent years, deep learning has greatly promoted the development of detection algorithms, but it requires a lot of markup data. In addition, the current common full-mark method has a large amount of work, a high degree of repetition and complex operation steps. In view of this situation, we combine the target detection technology with the whole marking method to design a new auxiliary marking method. We performed a labeling comparison experiment on the universal data set Pascal voc2012/2007. The results show that our method improves 2.5 times in efficiency with consistent accuracy. Based on the Common datasets, we also conducted experiments on the Chang’E series lunar CCD datasets, which also showed good performance and improved 2.8 times compared with the full mark indicates that our method can be applied to both general data sets and private data sets and also provides a research direction for the current auxiliary marking methods.