Research map / 3D vision / Depth and stereo
Monocular Depth Estimation Using Whole Strip Masking and Reliability-Based Refinement
Minhyeok Heo, Jaehan Lee, Kyung-Rae Kim, Han-Ul Kim, Chang-Su Kim
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 |
|---|---|---|---|
| replace the WSM layers with an equal-size square convolution (Equivalent decoder) vs WSM decoder (ablation) | ResNet-50 depth network · NYU Depth V2 (network model comparison) | rel: 0.150 -> 0.141 | Improved |
| use WSM layers instead of a 7x7 convolution (Inception decoder) (ablation) | ResNet-50 depth network · NYU Depth V2 (network model comparison) | rel: 0.148 -> 0.141 | Improved |
| use WSM decoder instead of UpProj decoder (ablation) | ResNet-50 depth network · NYU Depth V2 (network model comparison) | rms: 0.596 -> 0.582 | Improved |
| use WSM decoder instead of Deconv decoder (ablation) | ResNet-50 depth network · NYU Depth V2 (network model comparison) | rel: 0.152 -> 0.141 | Improved |
| use deeper backbone ResNet-50 instead of AlexNet (FC decoder) (ablation) | depth network with FC decoder · NYU Depth V2 (network model comparison) | rel: 0.215 -> 0.160 | Improved |
| use ResNet-50 instead of VGGNet-16 with Deconv decoder (ablation) | depth network with Deconv decoder · NYU Depth V2 (network model comparison) | rel: 0.194 -> 0.152 | Improved |
| use Deconv decoder instead of FC decoder with AlexNet (ablation) | AlexNet encoder · NYU Depth V2 (network model comparison) | rms: 0.833 -> 0.842 | Hurt |
| add reliability-based CRF refinement to WSM (ablation) | WSM · NYU Depth V2 (state-of-the-art comparison) | rel: 0.141 -> 0.135 | Improved |
| add reliability-based CRF refinement to WSM (ablation) | WSM · NYU Depth V2 (state-of-the-art comparison) | rms: 0.582 -> 0.571 | Improved |
| use reliability alpha in refinement instead of fixing alpha to 1 (ablation) | WSM-Ref · NYU Depth V2 (refinement analysis) | average error decrease: error decreases by up to 2.9% | Improved |
| add reliability-based refinement to WSM (ablation) | WSM · Make3D (C1 criterion) | rel: 0.185 -> 0.171 | Improved |
| add reliability-based refinement to WSM (ablation) | WSM · Make3D (C1 criterion) | rms: 4.85 -> 4.46 | Improved |
Closest papers by meaning
- Learning Depth from Single Monocular Images Using Deep Convolutional Neural Fields (TPAMI 2016 · similarity 0.90)
- Revisiting Gradient-Based Uncertainty for Monocular Depth Estimation (TPAMI 2025 · similarity 0.90)
- Unsupervised Monocular Depth Estimation With Left-Right Consistency (CVPR 2017 · similarity 0.89)
- Deep Digging into the Generalization of Self-Supervised Monocular Depth Estimation (AAAI 2023 · similarity 0.89)
- Monocular Depth Estimation Using Multi-Scale Continuous CRFs as Sequential Deep Networks (TPAMI 2018 · 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.
