Research map / Vision-language and multimodal / Human and embodied modeling
We Are Humor Beings: Understanding and Predicting Visual Humor
Check 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 combined Embedding, Cardinality, and Location scene features instead of the average-prediction baseline | Funniness-score SVR · AVH test | Avg. Rel. Err.: 0.3151 -> 0.2400 | Helped |
| Use Location features only instead of the combined features (ablation) | Funniness-score SVR · AVH test | Avg. Rel. Err.: 0.2400 -> 0.2400 | No clear effect |
| Use Cardinality features only instead of the combined features (ablation) | Funniness-score SVR · AVH test | Avg. Rel. Err.: 0.2400 -> 0.2450 | Hurt |
| Use Embedding features only instead of the combined features (ablation) | Funniness-score SVR · AVH test | Avg. Rel. Err.: 0.2400 -> 0.2516 | Hurt |
| Use Location rather than Cardinality features (ablation) | Funniness-score SVR · AVH test | Avg. Rel. Err.: 0.2450 -> 0.2400 | Helped |
| Add skip-connections that use predictions on other objects (ablation) | Object replacement detector · FOR funny to unfunny | performance | No clear effect |
| Use the learned detector rather than the strongest prior baseline for identifying objects to replace | Object replacement detector · FOR funny to unfunny | Avg. Cl. Acc.: 73.13% -> 74.45% | Helped |
| Use the learned detector rather than the do-not-replace prior | Object replacement detector · FOR funny to unfunny | Acc.: 79.86% -> 74.74% | Hurt |
| Use the learned replacer classifier rather than the prior baseline | Object replacer classifier · FOR funny to unfunny | Top-5 accuracy: 24.53% -> 29.65% | Helped |
| Replace humorous objects with the model's choices | Funny-to-unfunny pipeline · Funny-to-unfunny human evaluation | funniness: 2.69 -> 1.64 | Helped |
| Transform unfunny scenes using the model's object replacement pipeline | Unfunny-to-funny pipeline · Unfunny-to-funny human evaluation | funniness: 2.14 | Helped |
Closest papers by meaning
- HUMORCHAIN: Theory-Guided Multi-Stage Reasoning for Interpretable Multimodal Humor Generation (CVPR 2026 · similarity 0.89)
- Cracking the Code of Juxtaposition: Can AI Models Understand the Humorous Contradictions (NeurIPS 2024 · similarity 0.88)
- From Perception to Punchline: Empowering VLM with the Art of In-the-wild Memes (NeurIPS 2026 · similarity 0.87)
- Talk Funny! A Large-Scale Humor Response Dataset with Chain-of-Humor Interpretation (AAAI 2024 · similarity 0.85)
- Humor in AI: Massive Scale Crowd-Sourced Preferences and Benchmarks for Cartoon Captioning (NeurIPS 2024 · similarity 0.85)
- Visual Riddles: a Commonsense and World Knowledge Challenge for Large Vision and Language Models (NeurIPS 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.