Research map / Vision-language and multimodal / Human and embodied modeling
Automatic Attribute Discovery with Neural Activations
Sirion Vittayakorn, Takayuki Umeda, Kazuhiko Murasaki, Kyoko Sudo, Takayuki Okatani, Kota Yamaguchi
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 confidence-based resampling and all-activation classifier (ablation) | Pre-trained CaffeNet · Etsy visualness vs human | Spearman correlation: 0.637 -> 0.717 | Improved |
| add confidence-based resampling (ablation) | Pre-trained CaffeNet · Etsy visualness vs human | Pearson correlation: 0.737 -> 0.799 | Improved |
| add resampling (ablation) | Attribute-tuned CNN · Etsy visualness vs human | Spearman correlation: 0.565 -> 0.721 | Improved |
| add resampling (ablation) | Category-tuned CNN · Etsy visualness vs human | Spearman correlation: 0.684 -> 0.704 | Improved |
| use only 100 prime units (initial classifier) instead of 9,568-dim random classifier | Pre-trained CaffeNet · Etsy visualness vs human | Spearman correlation: 0.637 -> 0.663 | Improved |
| use 100 prime units instead of random baseline | Category-tuned CNN · Etsy visualness vs human | Spearman correlation: 0.684 -> 0.480 | Hurt |
| fine-tune CNN on noisy attribute words | CaffeNet · Etsy visualness vs human | Pearson correlation: 0.799 -> 0.782 | Hurt |
| use language prior instead of visual features | visualness estimation · Etsy visualness vs human | Pearson correlation: 0.799 -> 0.139 | Hurt |
Closest papers by meaning
- Automatic Discovery, Association Estimation and Learning of Semantic Attributes for a Thousand Categories (CVPR 2017 · similarity 0.89)
- Learning Concise and Descriptive Attributes for Visual Recognition (ICCV 2023 · similarity 0.87)
- Unsupervised Learning of Discriminative Attributes and Visual Representations (CVPR 2016 · similarity 0.87)
- Learning Deep Representations of Fine-Grained Visual Descriptions (CVPR 2016 · similarity 0.86)
- Learning Attributes Equals Multi-Source Domain Generalization (CVPR 2016 · similarity 0.86)
- Automated Detection of Visual Attribute Reliance with a Self-Reflective Agent (NeurIPS 2025 · similarity 0.84)
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.
