Research map / Vision-language and multimodal / Human and embodied modeling
Recognizing Material Properties from Images
Gabriel Schwartz, Ko Nishino
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 more discovered attributes beyond M=30 (ablation) | Discovered attributes + SVM · FMD | accuracy: less than 0.1% gain past M=30 | No clear effect |
| remove attribute layers from MAC-CNN (ablation) | MAC-CNN · Material database | average material category accuracy | No clear effect |
| constrain a top layer before softmax to output previously discovered attributes | Material recognition CNN · Attribute prediction | mean average error: 0.2 vs 0.08 | Hurt |
| add attribute constraint layer before softmax | Material recognition CNN · Material database | material accuracy | No clear effect |
| increase training set size for attribute discovery (ablation) | Discovered attributes · FMD | accuracy | No clear effect |
| add attribute probabilities to material probabilities as features (ablation) | Novel category linear SVM · Held-out material categories | novel category recognition accuracy | Improved |
| use local patches only instead of context | Material recognition · FMD | accuracy: 57.1% to 42.6% | Hurt |
Closest papers by meaning
- Hierarchical Material Recognition from Local Appearance (ICCV 2025 · similarity 0.88)
- Deep Filter Banks for Texture Recognition, Description, and Segmentation (IJCV 2016 · similarity 0.87)
- Connecting Look and Feel: Associating the Visual and Tactile Properties of Physical Materials (CVPR 2017 · similarity 0.87)
- Physical Property Understanding from Language-Embedded Feature Fields (CVPR 2024 · similarity 0.85)
- Infusing Synthetic Data with Real-World Patterns for Zero-Shot Material State Segmentation (NeurIPS 2024 · similarity 0.85)
- MaRI: Material Retrieval Integration across Domains (CVPR 2025 · 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.
