Research map / Vision-language and multimodal / CLIP and zero-shot
Exploring CLIP for Assessing the Look and Feel of Images
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 paired antonym prompts rather than a single prompt (ablation) | CLIP-IQA · KonIQ-10k and LIVE-itW | SROCC and PLCC | Helped |
| remove positional embeddings rather than retaining vanilla embeddings (ablation) | CLIP-IQA with ResNet-50 · KonIQ-10k | SROCC: 0.383 -> 0.695 | Helped |
| remove positional embeddings rather than retaining vanilla embeddings (ablation) | CLIP-IQA with ResNet-50 · KonIQ-10k | PLCC: 0.429 -> 0.727 | Helped |
| remove positional embeddings rather than interpolating them (ablation) | CLIP-IQA with ResNet-50 · KonIQ-10k | SROCC: 0.682 -> 0.695 | Helped |
| use the “[text] photo.” template rather than “A photo of [text].” (ablation) | CLIP-IQA · KonIQ-10k | SROCC: 0.116 -> 0.695 | Helped |
| use the “[text] photo.” template rather than “There is [text] in the photo” (ablation) | CLIP-IQA · KonIQ-10k | SROCC: 0.214 -> 0.695 | Helped |
| use Good/Bad rather than High quality/Low quality (ablation) | CLIP-IQA · KonIQ-10k | SROCC: 0.537 -> 0.695 | Helped |
| use Good/Bad rather than High definition/Low definition (ablation) | CLIP-IQA · KonIQ-10k | SROCC: 0.592 -> 0.695 | Helped |
| use ResNet-50 rather than VIT-B/32 with positional embedding removed (ablation) | CLIP-IQA · KonIQ-10k | SROCC: 0.391 -> 0.695 | Helped |
| finetune prompts with CoOp (ablation) | CLIP-IQA · TID2013 | SROCC: 0.51 -> 0.632 | Helped |
| finetune prompts with CoOp (ablation) | CLIP-IQA · TID2013 | PLCC: 0.571 -> 0.692 | Helped |
| use antonym prompts for fine-grained and abstract attributes without task-specific training | CLIP-IQA · synthetic quality attributes and AVA abstract attributes | attribute assessment consistency | Helped |
Closest papers by meaning
- Diffusion Feedback Helps CLIP See Better (ICLR 2025 · similarity 0.90)
- Q-Bench: A Benchmark for General-Purpose Foundation Models on Low-level Vision (ICLR 2024 · similarity 0.89)
- Demystifying CLIP Data (ICLR 2024 · similarity 0.88)
- The Many Senses of Visual Similarity: A Text-Prompted Image Perceptual Metric (NeurIPS 2026 · similarity 0.87)
- Post-pre-training for Modality Alignment in Vision-Language Foundation Models (CVPR 2025 · similarity 0.87)
- Probabilistic Prompt Adaptation for Unified Image Aesthetics and Quality Assessment (CVPR 2026 · similarity 0.87)
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.