Crater size-frequency distributions extracted from the crater database of La Grassa et al. (2025). Error bars (1σ uncertainty) are given according to Poisson statistics. Although the AI-based approach of identifying craters is spatially dependent, at the low latitudes the Chang’e-5 and 6 units are located, the false positives are minor (<3%), the recall rate is about 85%, and true positives reach at least 97% (La Grassa et al. 2025). As a result, the AI-based approach for crater identification yields consistent and reliable results for both units, recognising at least a similar fraction of the actual crater population.