mlsanityData to 14 Oct 2026

Deconstructing Self-Supervised Monocular Reconstruction: The Design Decisions that Matter

arXiv 2208.01489 ↗First posted 2 Aug 2022cs.CV cs.CG cs.LGcited by 18

Not on the map. Why?

Abstract (arXiv)

This paper presents an open and comprehensive framework to systematically evaluate state-of-the-art contributions to self-supervised monocular depth estimation. This includes pretraining, backbone, architectural design choices and loss functions. Many papers in this field claim novelty in either architecture design or loss formulation. However, simply updating the backbone of historical systems results in relative improvements of 25%, allowing them to outperform the majority of existing systems. A systematic evaluation of papers in this field was not straightforward. The need to compare like-with-like in previous papers means that longstanding errors in the evaluation protocol are ubiquitous in the field. It is likely that many papers were not only optimized for particular datasets, but also for errors in the data and evaluation criteria. To aid future research in this area, we release a modular codebase (https://github.com/jspenmar/monodepth_benchmark), allowing for easy evaluation of alternate design decisions against corrected data and evaluation criteria. We re-implement, validate and re-evaluate 16 state-of-the-art contributions and introduce a new dataset (SYNS-Patches) containing dense outdoor depth maps in a variety of both natural and urban scenes. This allows for the computation of informative metrics in complex regions such as depth boundaries.

Key facts. Cited by 18 papers in the map (data through 14 Oct 2026). Builds on 45 of its 58 references that are in the map. First posted on arXiv 2 Aug 2022.

Builds on

45 of 58 references.

References · 45

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,442
  3. 2017Squeeze-and-Excitation Networkscited by 4,789
  4. 2022A ConvNet for the 2020scited by 3,579
  5. 2016Aggregated Residual Transformations for Deep Neural Networkscited by 2,546
  6. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,352
  7. 2019Searching for MobileNetV3cited by 1,994
  8. 2015Spatial Transformer Networkscited by 1,429
  9. 2019Deep High-Resolution Representation Learning for Visual Recognitioncited by 1,420
  10. 2016Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Networkcited by 1,204
  11. 2015A Large Dataset to Train Convolutional Networks for Disparity, Optical Flow, and Scene Flow Estimationcited by 881
  12. 2016Unsupervised Monocular Depth Estimation with Left-Right Consistencycited by 820
33 more
  1. 2017Unsupervised Learning of Depth and Ego-Motion from Videocited by 782
  2. 2016Direct Sparse Odometrycited by 485
  3. 2014Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-Scale Convolutional Architecturecited by 483
  4. 20193D Packing for Self-Supervised Monocular Depth Estimationcited by 439
  5. 2016Deeper Depth Prediction with Fully Convolutional Residual Networkscited by 403
  6. 2016Unsupervised CNN for Single View Depth Estimation: Geometry to the Rescuecited by 368
  7. 2017Sparsity Invariant CNNscited by 326
  8. 2019Unsupervised Scale-consistent Depth and Ego-motion Learning from Monocular Videocited by 244
  9. 2019Billion-scale semi-supervised learning for image classificationcited by 232
  10. 2020D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometrycited by 206
  11. 2020On the uncertainty of self-supervised monocular depth estimationcited by 182
  12. 2018Unsupervised Learning of Depth and Ego-Motion from Monocular Video Using 3D Geometric Constraintscited by 179
  13. 2018Unsupervised Learning of Monocular Depth Estimation and Visual Odometry with Deep Feature Reconstructioncited by 159
  14. 2017Learning Depth from Monocular Videos using Direct Methodscited by 154
  15. 2020Feature-metric Loss for Self-supervised Learning of Depth and Egomotioncited by 140
  16. 2018Evaluation of CNN-based Single-Image Depth Estimation Methodscited by 138
  17. 2019Self-Supervised Monocular Depth Hintscited by 127
  18. 2020Self-supervised Monocular Trained Depth Estimation using Self-attention and Discrete Disparity Volumecited by 116
  19. 2020HR-Depth: High Resolution Self-Supervised Monocular Depth Estimationcited by 115
  20. 2017Semi-Supervised Deep Learning for Monocular Depth Map Predictioncited by 109
  21. 2018Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometrycited by 85
  22. 2021Self-Supervised Monocular Depth Estimation with Internal Feature Fusioncited by 67
  23. 2018SuperDepth: Self-Supervised, Super-Resolved Monocular Depth Estimationcited by 64
  24. 2019Learning monocular depth estimation infusing traditional stereo knowledgecited by 64
  25. 2021Channel-Wise Attention-Based Network for Self-Supervised Monocular Depth Estimationcited by 63
  26. 2020DeFeat-Net: General Monocular Depth via Simultaneous Unsupervised Representation Learningcited by 60
  27. 2018SDCNet: Video Prediction Using Spatially-Displaced Convolutioncited by 50
  28. 2020Forget About the LiDAR: Self-Supervised Depth Estimators with MED Probability Volumescited by 50
  29. 2017Stereo DSO: Large-Scale Direct Sparse Visual Odometry with Stereo Camerascited by 45
  30. 2018Single View Stereo Matchingcited by 39
  31. 2018Learning monocular depth estimation with unsupervised trinocular assumptionscited by 36
  32. 2021PLADE-Net: Towards Pixel-Level Accuracy for Self-Supervised Single-View Depth Estimation with Neural Positional Encoding and Distilled Matting Losscited by 31
  33. 2022From 2D to 3D: Re-thinking Benchmarking of Monocular Depth Predictioncited by 19

Built on by

Cited by year

Citing papers per year: 2022: 1; 2023: 5; 2024: 8; 2025: 4.

18 papers.

Sort:

By area:

  1. 2024Depth Pro: Sharp Monocular Metric Depth in Less Than a Secondcited by 404
  2. 2023Kick Back & Relax: Learning to Reconstruct the World by Watching SlowTVcited by 21
  3. 2024DepthCues: Evaluating Monocular Depth Perception in Large Vision Modelscited by 19
  4. 2022The Monocular Depth Estimation Challengecited by 13
  5. 2023The Second Monocular Depth Estimation Challengecited by 12
  6. 2024The Third Monocular Depth Estimation Challengecited by 10
  7. 2023Scalable Semantic 3D Mapping of Coral Reefs with Deep Learningcited by 8
  8. 2025The Fourth Monocular Depth Estimation Challengecited by 8
  9. 2024Kick Back & Relax++: Scaling Beyond Ground-Truth Depth with SlowTV & CribsTVcited by 7
  10. 2025Position: Machine Learning Conferences Should Establish a "Refutations and Critiques" Trackcited by 7
  11. 2023Equivariant Single View Pose Prediction Via Induced and Restricted Representationscited by 6
  12. 2025RPG360: Robust 360 Depth Estimation with Perspective Foundation Models and Graph Optimizationcited by 4
6 more
  1. 2024UnDIVE: Generalized Underwater Video Enhancement Using Generative Priorscited by 3
  2. 2024BaseBoostDepth: Exploiting Larger Baselines For Self-supervised Monocular Depth Estimationcited by 2
  3. 2024NimbleD: Enhancing Self-supervised Monocular Depth Estimation with Pseudo-labels and Large-scale Video Pre-trainingcited by 2
  4. 2023HQDec: Self-Supervised Monocular Depth Estimation Based on a High-Quality Decodercited by 1
  5. 2025Towards Sharper Object Boundaries in Self-Supervised Depth Estimationcited by 1
  6. 2024EDADepth: Enhanced Data Augmentation for Monocular Depth Estimationcited by 0