Skip to main content
[Preprint]. 2024 Apr 3:2023.04.28.538703. Originally published 2023 Apr 28. [Version 2] doi: 10.1101/2023.04.28.538703

Extended Data Figure 5: Unlabeled frames improve pose estimation in CRIM13 dataset.

Extended Data Figure 5:

Models in panel A use 800 training frames. Remaining conventions as in Fig. 4. Also see Supplementary Video 7.