Research map / Architectures and efficiency / Attention and transformers
GuideFormer: Transformers for Image Guided Depth Completion
Kyeongha Rho, Jinsung Ha, Youngjung Kim
Check 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 |
|---|---|---|---|
| use a Transformer encoder-decoder instead of a CNN encoder-decoder (ablation) | GuideFormer · KITTI validation set | RMSE: 772.78 -> 765.38 | Improved |
| replace the depth-and-color concatenation with pre-guide GAM (ablation) | GuideFormer · KITTI validation set, parallel guidance | RMSE: 768.47 -> 756.15 | Improved |
| replace the depth-and-color concatenation with pre-guide GAM (ablation) | GuideFormer · KITTI validation set, sequential guidance | RMSE: 765.38 -> 754.17 | Improved |
| use pre-guide instead of post-guide GAM (ablation) | GuideFormer · KITTI validation set, parallel guidance | RMSE: 759.98 -> 756.15 | Improved |
| use pre-guide instead of post-guide GAM (ablation) | GuideFormer · KITTI validation set, sequential guidance | RMSE: 757.52 -> 754.17 | Improved |
| add reverse guided-attention to make GAM bidirectional (ablation) | GuideFormer · KITTI validation set, parallel guidance | RMSE: 756.15 -> 755.09 | Improved |
| process color and depth features sequentially rather than in parallel (ablation) | GuideFormer with pre-guide GAM · KITTI validation set | RMSE: 756.15 -> 754.17 | Improved |
| use dual branches instead of a single branch (ablation) | GuideFormer · KITTI validation set | RMSE: 867.94 -> 754.17 | Improved |
| add depth fusion of the color-branch and depth-branch depth predictions (ablation) | GuideFormer dual-branch model · KITTI validation set, parallel guidance | RMSE: 759.61 -> 756.15 | Improved |
| add depth fusion of the color-branch and depth-branch depth predictions (ablation) | GuideFormer dual-branch model · KITTI validation set, sequential guidance | RMSE: 758.26 -> 754.17 | Improved |
| use GuideFormer instead of PENet | KITTI depth completion · KITTI test set | RMSE: 730.08 -> 721.48 | Improved |
Closest papers by meaning
- CompletionFormer: Depth Completion With Convolutions and Vision Transformers (CVPR 2023 · similarity 0.90)
- RigNet: Repetitive Image Guided Network for Depth Completion (ECCV 2022 · similarity 0.89)
- Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion (CVPR 2025 · similarity 0.88)
- DeCoTR: Enhancing Depth Completion with 2D and 3D Attentions (CVPR 2024 · similarity 0.86)
- Improving Depth Completion via Depth Feature Upsampling (CVPR 2024 · similarity 0.85)
- DFormerv2: Geometry Self-Attention for RGBD Semantic Segmentation (CVPR 2025 · 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.
