Research map / Vision-language and multimodal / CLIP and zero-shot
CLIP-Event: Connecting Text and Images With Event Structures
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 |
|---|---|---|---|
| Pretrain CLIP with event-structure contrastive learning and graph alignment | CLIP ViT-B/32 · M2E2 zero-shot | Event F1: 42.6 -> 48.1 | Helped |
| Pretrain CLIP with event-structure contrastive learning and graph alignment | CLIP ViT-B/32 · M2E2 zero-shot | Argument F1: 11.1 -> 14.8 | Helped |
| Pretrain CLIP with event-structure contrastive learning and graph alignment | CLIP ViT-B/32 · VOANews text-to-image retrieval | Recall@1: 23.5 -> 27.5 | Helped |
| Pretrain CLIP with event-structure contrastive learning and graph alignment | CLIP ViT-B/32 · VCR | Rationale F1: 47.2 -> 49.2 | Helped |
| Add event graph alignment via optimal transport (ablation) | CLIP-Event · M2E2 zero-shot | Argument F1: 11.9 -> 14.8 | Helped |
| Add event graph alignment via optimal transport (ablation) | CLIP-Event · VOANews text-to-image retrieval | Recall@1: 25.5 -> 27.5 | Helped |
| Add event graph alignment via optimal transport (ablation) | CLIP-Event · VCR | Answer F1: 52.0 -> 52.4 | Helped |
| Add event graph alignment via optimal transport (ablation) | CLIP-Event · VisualCOMET | Accuracy@50: 21.1 -> 22.4 | Helped |
| Use GPT-3-generated prompts instead of the full prompt-function setup (ablation) | CLIP-Event · M2E2 zero-shot | Event F1: 48.1 -> 47.3 | Hurt |
| Use only single-template prompts instead of the full prompt-function setup (ablation) | CLIP-Event · M2E2 zero-shot | Event F1: 48.1 -> 44.4 | Hurt |
| Use only composed-template prompts instead of the full prompt-function setup (ablation) | CLIP-Event · M2E2 zero-shot | Event F1: 48.1 -> 46.3 | Hurt |
| Use only continuous prompts instead of the full prompt-function setup (ablation) | CLIP-Event · M2E2 zero-shot | Event F1: 48.1 -> 46.5 | Hurt |
Closest papers by meaning
- EventGPT: Event Stream Understanding with Multimodal Large Language Models (CVPR 2025 · similarity 0.88)
- EventLens: Event-Structure Reinforcement Learning for Video Understanding (NeurIPS 2026 · similarity 0.88)
- RE-VLM: Event-Augmented Vision-Language Model for Scene Understanding (CVPR 2026 · similarity 0.87)
- SuperCLIP: CLIP with Simple Classification Supervision (NeurIPS 2025 · similarity 0.87)
- OpenESS: Event-based Semantic Scene Understanding with Open Vocabularies (CVPR 2024 · similarity 0.87)
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · 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.