Fig. 8: Sample feature size and testing performance.

a Quantitative comparison of testing performance in terms of feature size, and cross-domain generalization result of (b) the baseline model (21 M) and (c) the proposed RNN (21 M). d Qualitative comparison of a volumetric reconstruction for different cases of training and testing conditions in terms of feature size (see Table S1 in Supplementary Information (SI) Section S1 for more details)