section 2 "related work", sub-section "learning based methods", paragraphs 1 to 4, page 3
section 3.1 "learning to enhance reconstructions", paragraphs 1 and 3, page 4
section 3.2 "image enhancement", paragraphs 1 to 5, page 4
sub-section "stereo loss", page 5
section 4 "evaluation", paragraphs 1 and 2, page 7 - single camera dataset
section 4.1 "volumetric capture", paragraphs 1 and 2
sub-section "full body capture (multi-view)", paragraph3, page 8 - viewpoint changes
section 4.2 "qualitative results", sub-section "upper body results (single view)", page 8
section 4.2 "qualitative results", sub-section "viewpoint robustness", page 8
section 4.3 "ablation study", sub-section "segmentation mask", page 9
section 5 "real-time free viewpoint neural re-rendering", page 9
section 6 "discussion, limitations and future work", last paragraph, page 12
figures 1,3,4,5,7,8,12,17