Research map / Architectures and efficiency / CNN design
SDC - Stacked Dilated Convolution: A Unified Descriptor Network for Dense Matching Tasks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use the SDC descriptor instead of the original ELAS descriptor | ELAS · KITTI noc all stereo | EPE: 1.38 -> 1.13 | Helped |
| Use the SDC descriptor instead of the original ELAS descriptor | ELAS · ETH3D all stereo | EPE: 1.02 -> 0.64 | Helped |
| Use the SDC descriptor instead of CENSUS | SGM · KITTI noc fg stereo | >3px: 6.46 -> 7.70 | Hurt |
| Use the SDC descriptor instead of SIFT | FlowFields++ · KITTI noc all optical flow, matching | EPE: 12.48 -> 6.19 | Helped |
| Use the SDC descriptor instead of SIFT | FlowFields++ · HD1K optical flow, matching | EPE: 12.99 -> 7.48 | Helped |
| Use the SDC descriptor instead of SIFT | CPM · Sintel all optical flow | EPE: 1.28 -> 1.50 | Hurt |
| Use the SDC descriptor instead of SIFTFlow | SceneFlowFields · KITTI D1 noc all scene flow, matching | >3px: 10.91 -> 5.53 | Helped |
| Use a compact 4-block network with 3 shared-weight parallel dilated convolutions (ablation) | SDC descriptor · Patch-triplet test | accuracy: 97.2~\% -> 96.0~\% | Hurt |
Closest papers by meaning
- Learning for Disparity Estimation Through Feature Constancy (CVPR 2018 · similarity 0.88)
- Efficient Deep Learning for Stereo Matching (CVPR 2016 · similarity 0.87)
- Densely Connected Convolutional Networks (CVPR 2017 · similarity 0.86)
- Local Similarity Pattern and Cost Self-Reassembling for Deep Stereo Matching Networks (AAAI 2022 · similarity 0.86)
- Unifying Feature and Cost Aggregation with Transformers for Semantic and Visual Correspondence (ICLR 2024 · similarity 0.85)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.85)
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.