The Fourth Monocular Depth Estimation Challenge

Anton Obukhov, Matteo Poggi, Fabio Tosi, Ripudaman Singh Arora, Jaime Spencer, Chris Russell, Simon Hadfield, Richard Bowden, Shuaihang Wang, Zhenxin Ma, Weijie Chen, Baobei Xu, Fengyu Sun, Di Xie, Jiang Zhu, Mykola Lavreniuk, Haining Guan, Qun Wu, Yupei Zeng, Chao Lu et al. (37 additional authors not shown)

Abstract: This paper discusses the results of the fourth edition of the Monocular Depth Estimation Challenge (MDEC), a challenge focusing on zero-shot generalization to the SYNS-Patches benchmark, a dataset featuring challenging environments in both natural and indoor settings. In this edition, we switched the evaluation protocol to perform least-squares rather than median alignment. The challenge received a total of 24 submissions outperforming the baselines on the test set: 10 among them submitted a report describing their approach, with a diffuse use of Depth Anything v2 at the core of most methods. The challenge winners improved 3D F-Score performance over the previous edition winners from 22.58% to 23.05%.