Application of Deep Learning in Lunar Volcanic Dome Identification

Lunar domes have always been one of the important windows to understand lunar volcanic activity, however traditional identification methods for geological domes are expensive, so this study attempts to establish an automatic identification method for lunar volcanic domes. Given that no previous research in this area has attempted to automate the identification of lunar volcanic domes, our team attempted to automate the process for the first time. To achieve the purpose of this research, the researchers first obtained the dome coordinates from the list of known lunar domes and intercepted the data we needed from the corresponding coordinates on the CCD and DEM moon pictures. Subsequently, the researchers screened the data to find data with more obvious features and used these data to train 9 mainstream image recognition models and compared their accuracy rates to verify the feasibility of this study. Finally, the researchers counted the mAP and AP (IoU=0.5) of the nine models and found that the highest of them could reach 0.64 (mAP) and 0.74 (AP). Therefore, this study can conclude that an automated method for identifying lunar volcanic domes should be feasible.