Research map / Vision-language and multimodal / Human and embodied modeling
Proxy Learning of Visual Concepts of Fine Art Paintings from Styles through Language Models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use deep-proxy SVD instead of PCA or ESZSL with BERT style relations | Visual concept prediction on WikiArt · WikiArt eval, abstract | AUC: 0.8 -> 0.9; 0.7 -> 0.9 | Helped |
| Use GloVe rather than BERT style relations | LGT · WikiArt eval-VAL, planar | AUC: 0.38 -> 0.77 | Helped |
| Use GloVe rather than BERT style relations | Deep-proxy SVD · WikiArt eval-VAL, planar | AUC: 0.73 -> 0.78 | Helped |
| Use deep-proxy SVD instead of LGT with BERT style relations | Visual concept prediction · WikiArt eval-VAL, planar | AUC: 0.38 -> 0.73 | Helped |
| Replace the Plain category-attribute matrix with the SVD-transformed matrix (ablation) | Deep-proxy · WikiArt eval | AUC | Helped |
| Add a learnable offset to the category-attribute matrix (ablation) | Deep-proxy · WikiArt eval | AUC | Hurt |
| Fine-tune the convolutional layers instead of training only the fully connected layers (ablation) | Deep-proxy VGG-16 · WikiArt | AUC | Hurt |
| Use deep-proxy SVD rather than LGT with ground-truth style relations | Visual concept prediction · WikiArt eval, diagonal, parallel, straight, and gestural | AUC | Hurt |
| Use deep-proxy SVD rather than LGT with language-model style relations | Visual concept prediction · WikiArt eval-VAL, top 15 visual concepts | AUC | Mixed |
| Use deep-proxy SVD rather than LGT with language-model style relations | Visual concept prediction · WikiArt, language-model quality comparison | AUC correlation with relation quality | Helped |
| Use Offset rather than LGT when the category-attribute matrix is sparse (ablation) | Proxy learning on SUN · SUN validation | mean AP | Helped |
| Use deep-proxy SVD with the CUB category-attribute matrix (ablation) | Deep-proxy · CUB | mean AP | Hurt |
Closest papers by meaning
- APDDv2: Aesthetics of Paintings and Drawings Dataset with Artist Labeled Scores and Comments (NeurIPS 2024 · similarity 0.87)
- Bridging Cognitive Gap: Hierarchical Description Learning for Artistic Image Aesthetics Assessment (AAAI 2026 · similarity 0.86)
- AACP: Aesthetics Assessment of Children’s Paintings Based on Self-Supervised Learning (AAAI 2024 · similarity 0.85)
- ArtBank: Artistic Style Transfer with Pre-trained Diffusion Model and Implicit Style Prompt Bank (AAAI 2024 · similarity 0.85)
- ConceptBed: Evaluating Concept Learning Abilities of Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.85)
- Stylebreeder: Exploring and Democratizing Artistic Styles through Text-to-Image Models (NeurIPS 2024 · 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.