Research map / 3D vision / Depth and stereo
Differential Angular Imaging for Material Recognition
Jia Xue, Hang Zhang, Kristin Dana, Ko Nishino
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 |
|---|---|---|---|
| Add a differential image stream to single-view CNN recognition | GTOS material recognition · GTOS, 5 splits | accuracy: 74.3 ±2.8 -> 79.4 ±3.4 | Improved |
| Use the differential image rather than the second viewpoint image (ablation) | Single-view DAIN (Sum) · GTOS, 5 splits | accuracy: 77.5 ±2.7 -> 79.4 ±3.4 | Improved |
| Use the hybrid fusion that combines feature maps and prediction outputs (ablation) | Two-stream DAIN · GTOS, 5 splits | accuracy: 77.0 ±2.5 -> 79.4 ±3.4 | Improved |
| Combine streams at the intermediate layer only, without hybrid prediction-level fusion (ablation) | Two-stream DAIN · GTOS, 5 splits | accuracy: 79.4 ±3.4 -> 74.8 ±3.4 | Hurt |
| Replace Sum feature-map fusion with Max fusion (ablation) | Single-view DAIN · GTOS, 5 splits | accuracy: 79.4 ±3.4 -> 79.0 ±1.8 | No clear effect |
| Add multiview inputs to single-view DAIN | DAIN · GTOS, 5 splits | accuracy: 79.4 ±3.4 -> 81.2 ±1.7 | Improved |
| Use 3D filter plus pooling instead of voting to combine multiview DAIN (ablation) | Multiview DAIN · GTOS, 5 splits | accuracy: 80.0 ±2.1 -> 81.1 ±1.5 | Improved |
| Replace VGG-M with ResNet-50 (ablation) | Single-view DAIN (Sum) · GTOS, split 1 | accuracy: 77.5% -> 83.0% | Improved |
Closest papers by meaning
- Differential Viewpoints for Ground Terrain Material Recognition (TPAMI 2020 · similarity 0.97)
- Deep Material Recognition in Light-Fields via Disentanglement of Spatial and Angular Information (ECCV 2020 · similarity 0.89)
- Hierarchical Material Recognition from Local Appearance (ICCV 2025 · similarity 0.87)
- Infusing Synthetic Data with Real-World Patterns for Zero-Shot Material State Segmentation (NeurIPS 2024 · similarity 0.85)
- A 4D Light-Field Dataset and CNN Architectures for Material Recognition (ECCV 2016 · similarity 0.85)
- Material Palette: Extraction of Materials from a Single Image (CVPR 2024 · similarity 0.83)
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.
