Research map / 3D vision / Depth and stereo
DustNet++: Deep Learning-Based Visual Regression for Dust Density Estimation
Andreas Michel, Martin Weinmann, Jannick Kuester, Faisal AlNasser, Tomás Gómez, Mark J. Falvey, Rainer Schmitz, Wolfgang Middelmann, Stefan Hinz
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 MBConv (ablation) | DustNet++ Swin-L · Meteodata dust | MAE: 13.74 -> 12.63 | Improved |
| add window attention to grid attention (ablation) | DustNet++ Swin-L · Meteodata dust | MAE: 17.94 -> 12.63 | Improved |
| add grid attention to window attention (ablation) | DustNet++ Swin-L · Meteodata dust | MAE: 15.62 -> 12.63 | Improved |
| add vanilla MSA alongside cross attention (ablation) | DustNet++ Swin-L · Meteodata dust | MAE: 18.93 -> 12.63 | Improved |
| add cross multi-axis attention to vanilla attention (ablation) | DustNet++ Swin-L · Meteodata dust | MAE: 15.46 -> 12.63 | Improved |
| use checkpoint wrapper to double batch size (ablation) | DustNet S ResNet101 · Meteodata dust | MAE: 19.27 -> 14.10 | Improved |
| use checkpoint wrapper to double batch size (ablation) | PixelFormer Swin-B · Meteodata dust | MAE: 21.53 -> 17.43 | Improved |
| use checkpoint wrapper for NewCRF Swin-B (ablation) | NewCRF Swin-B · Meteodata dust | MAE: 20.00 -> 20.83 | Hurt |
| use Swin-L backbone instead of ResNet101 | DustNet++ · Meteodata dust (single image) | MAE: 13.95 -> 12.63 | Improved |
| add a second input image (temporal) | DustNet++ Swin-L · Meteodata dust | MAE: 12.63 -> 11.73 | Improved |
| use DustNet++ instead of DustNet S | Swin-L · Meteodata dust (single image) | IoU: 0.793 -> 0.784 | Mixed |
| use DustNet++ instead of DustNet S for generalization | Swin-L · URDE RandomDataset_897 | IoU: 0.618 -> 0.631 | Improved |
Closest papers by meaning
- Dust-Mamba: An Efficient Dust Storm Detection Network with Multiple Data Sources (AAAI 2025 · similarity 0.86)
- AirRadar: Inferring Nationwide Air Quality in China with Deep Neural Networks (AAAI 2025 · similarity 0.81)
- MV-DUSt3R+: Single-Stage Scene Reconstruction from Sparse Views In 2 Seconds (CVPR 2025 · similarity 0.80)
- UrbanWaste: In-the-Bin Dataset for Waste Disposal Inspection with Multi-Granularity Hierarchical Labels (AAAI 2025 · similarity 0.79)
- MDFL: Multi-Domain Diffusion-Driven Feature Learning (AAAI 2024 · similarity 0.79)
- Tokenize Image Patches: Global Context Fusion for Effective Haze Removal in Large Images (CVPR 2025 · similarity 0.79)
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.
