Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Third Monocular Depth Estimation Challenge

Jaime Spencer, Fabio Tosi, Matteo Poggi, Ripudaman Singh Arora, Chris Russell, Simon Hadfield, Richard Bowden, GuangYuan Zhou, ZhengXin Li, Qiang Rao, YiPing Bao, Xiao Liu, Dohyeong Kim, Jinseong Kim, Myunghyun Kim, Mykola Lavreniuk, Rui Li, Qing Mao, Jiang Wu, Yu Zhu, Jinqiu Sun, Yanning Zhang, Suraj Patni, Aradhye Agarwal, Chetan Arora, Pihai Sun, Kui Jiang, Gang Wu, Jian Liu, Xianming Liu, Junjun Jiang, Xidan Zhang, Jianing Wei, Fangjun Wang, Zhiming Tan, Jiabao Wang, Albert Luginov, Muhammad Shahzad, Seyed Hosseini, Aleksander Trajcevski, James H. Elder

TL;DR

This paper reports the third Monocular Depth Estimation Challenge (MDEC), focusing on zero-shot generalization to the SYNS-Patches dataset across diverse natural and indoor scenes. It allows any supervision and leverages withheld test-ground-truth to stress generalization, with Depth Anything emerging as a dominant backbone across many top submissions. The results show substantial gains over the prior edition, culminating in a top F-score of 23.72 and highlighting the impact of large-scale pretraining on depth estimation. The study also analyzes qualitative trends, challenges in thin or non-Lambertian regions, and suggests directions for future work and datasets to further close remaining gaps in monocular depth estimation.

Abstract

This paper discusses the results of the third edition of the Monocular Depth Estimation Challenge (MDEC). The challenge focuses on zero-shot generalization to the challenging SYNS-Patches dataset, featuring complex scenes in natural and indoor settings. As with the previous edition, methods can use any form of supervision, i.e. supervised or self-supervised. The challenge received a total of 19 submissions outperforming the baseline on the test set: 10 among them submitted a report describing their approach, highlighting a diffused use of foundational models such as Depth Anything at the core of their method. The challenge winners drastically improved 3D F-Score performance, from 17.51% to 23.72%.

The Third Monocular Depth Estimation Challenge

TL;DR

This paper reports the third Monocular Depth Estimation Challenge (MDEC), focusing on zero-shot generalization to the SYNS-Patches dataset across diverse natural and indoor scenes. It allows any supervision and leverages withheld test-ground-truth to stress generalization, with Depth Anything emerging as a dominant backbone across many top submissions. The results show substantial gains over the prior edition, culminating in a top F-score of 23.72 and highlighting the impact of large-scale pretraining on depth estimation. The study also analyzes qualitative trends, challenges in thin or non-Lambertian regions, and suggests directions for future work and datasets to further close remaining gaps in monocular depth estimation.

Abstract

This paper discusses the results of the third edition of the Monocular Depth Estimation Challenge (MDEC). The challenge focuses on zero-shot generalization to the challenging SYNS-Patches dataset, featuring complex scenes in natural and indoor settings. As with the previous edition, methods can use any form of supervision, i.e. supervised or self-supervised. The challenge received a total of 19 submissions outperforming the baseline on the test set: 10 among them submitted a report describing their approach, highlighting a diffused use of foundational models such as Depth Anything at the core of their method. The challenge winners drastically improved 3D F-Score performance, from 17.51% to 23.72%.
Paper Structure (8 sections)

This paper contains 8 sections.

Table of Contents

  1. Introduction
  2. Related Work
  3. The Monocular Depth Estimation Challenge
  4. Challenge Submissions
  5. Results
  6. Quantitative Results
  7. Qualitative Results
  8. Conclusions & Future Work