Research map / 3D vision / Depth and stereo
A Study on the Generality of Neural Network Structures for Monocular Depth Estimation
Jinwoo Bae, Kyumin Hwang, Sunghoon Im
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ACM and FFD together (ablation) | MonoFormer baseline · KITTI Eigen test | Abs Rel: 0.113 -> 0.104 | Improved |
| add ACM only (ablation) | MonoFormer baseline · KITTI Eigen test | Abs Rel: 0.113 -> 0.113 | No clear effect |
| add FFD only (ablation) | MonoFormer baseline · KITTI Eigen test | Abs Rel: 0.113 -> 0.112 | Improved |
| use 4 transformer layers instead of 2 (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.148 -> 0.104 | Improved |
| use 4 transformer layers instead of 5 (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.111 -> 0.104 | Improved |
| use 4 transformer layers instead of 3 (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.112 -> 0.104 | Improved |
| replace CNN-Transformer encoder with ConvNeXt (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.104 -> 0.111 | Hurt |
| replace CNN-Transformer encoder with SLaK (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.104 -> 0.117 | Hurt |
| replace CNN-Transformer encoder with Twins (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.104 -> 0.125 | Hurt |
| replace CNN-Transformer encoder with plain ViT (ablation) | MonoFormer · KITTI Eigen test | Abs Rel: 0.104 -> 0.118 | Hurt |
| use ground-truth depth supervision instead of self-supervision | depth networks · KITTI Eigen test | Abs Rel: 0.104 -> 0.052 | Improved |
| use ConvNeXt (macro design mimicking attention) instead of other CNN encoders | self-supervised MDE · out-of-distribution and texture-shifted datasets | generalization | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- R-MSFM6 — R-MSFM: Recurrent Multi-Scale Feature Modulation for Monocular Depth Estimating (ICCV 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Deep Digging into the Generalization of Self-Supervised Monocular Depth Estimation (AAAI 2023 · similarity 0.97)
- Visualization of Convolutional Neural Networks for Monocular Depth Estimation (ICCV 2019 · similarity 0.90)
- Improving Monocular Depth Estimation by Leveraging Structural Awareness and Complementary Datasets (ECCV 2020 · similarity 0.89)
- Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement (ECCV 2018 · similarity 0.89)
- Depth from a Single Image by Harmonizing Overcomplete Local Network Predictions (NeurIPS 2016 · similarity 0.89)
- DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
