Research map / 3D vision / Depth and stereo
Reconstructing Dense Depth of Dark Scenes with Sparse LiDAR, Noisy Events, and Blurry RGB
Jianbo Cao, Yuqi Han, Siming Zheng, Bo Wang, Tong Guo, Jinli Suo
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 event-guided feature restoration (EFR) instead of a deblurrer (ablation) | ERU-Net · Night-RIDE (real-world) | RMSE: 4.215 -> 1.505 | Improved |
| add hybrid-guidance confidence module (HGCM) event guidance (ablation) | ERU-Net · Night-RIDE (real-world) | RMSE: 2.345 -> 1.505 | Improved |
| add EFR instead of deblurrer (ablation) | ERU-Net · Night-RIDE (real-world) | MAE: 1.682 -> 0.502 | Improved |
| add HGCM (ablation) | ERU-Net · Night-RIDE (real-world) | MAE: 0.892 -> 0.502 | Improved |
| add HGCM (ablation) | ERU-Net · Night-RIDE (real-world) | delta1: 98.1 -> 99.6 | Improved |
| add EFR instead of deblurrer (ablation) | ERU-Net · Night-RIDE (real-world) | delta1: 92.5 -> 99.6 | Improved |
| replace EVSSM+Mari-DC two-stage pipeline with end-to-end ERU-Net | ERU-Net · KITTI (simulated blur) | RMSE: 1093.50 -> 888.20 | Improved |
| replace EVSSM+Mari-DC with ERU-Net | ERU-Net · NYUv2 (500 sparse points) | RMSE: 3.902 -> 2.603 | Improved |
| use REFID instead of EFNet as event deblurring pre-processing | Mari-DC · Event-deblurring + depth completion (Table 2) | RMSE: 3.352 -> 3.145 | Improved |
| use REFID instead of EFNet pre-processing | CSPN · Event-deblurring + depth completion (Table 2) | RMSE: 6.090 -> 5.782 | Improved |
| replace restore-then-complete event pipeline (REFID+Mari-DC) with ERU-Net | ERU-Net · Event-deblurring + depth completion (Table 2) | RMSE: 3.145 -> 2.603 | Improved |
| add three-stage training with joint fine-tuning | ERU-Net · KITTI / NYUv2 | RMSE | Improved |
Closest papers by meaning
- Event-Driven Dynamic Scene Depth Completion (NeurIPS 2025 · similarity 0.90)
- EAG3R: Event-Augmented 3D Geometry Estimation for Dynamic and Extreme-Lighting Scenes (NeurIPS 2025 · similarity 0.86)
- DERD-Net: Learning Depth from Event-based Ray Densities (NeurIPS 2025 · similarity 0.86)
- Towards Robust Event-guided Low-Light Image Enhancement: A Large-Scale Real-World Event-Image Dataset and Novel Approach (CVPR 2024 · similarity 0.86)
- Bring Event into RGB and LiDAR: Hierarchical Visual-Motion Fusion for Scene Flow (CVPR 2024 · similarity 0.86)
- Gated Fields: Learning Scene Reconstruction from Gated Videos (CVPR 2024 · similarity 0.86)
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.
