Research map / Video / Video-language understanding
ViTAL‑X: Video-Text Alignment with Cross‑Modal Temporal Edits
Sethuraman T V, Savya Khosla, Onkar Susladkar, Aditi Tiwari, Seoung Wug Oh, Kushal Kafle, Joon-Young Lee, Derek Hoiem, Simon Jenni
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 sequence composition edit (ablation) | ViTAL-X · VideoComp | accuracy: 61.4 -> 67.8 | Improved |
| add sequence composition edit (ablation) | ViTAL-X · RTime | accuracy: 59.1 -> 65.3 | Improved |
| add clip reordering edit (ablation) | ViTAL-X · YouCook2 | accuracy: 46.5 -> 50.4 | Improved |
| add temporal cropping edit (ablation) | ViTAL-X · DiDeMo | accuracy: 52.5 -> 55.8 | Improved |
| add reversal edit (ablation) | ViTAL-X · Avg-Temp | accuracy: 58.4 -> 61.1 | Improved |
| add textual hard negatives (ablation) | ViTAL-X · Avg-Temp | accuracy: 58.8 -> 61.1 | Improved |
| use spatio-temporal adapter instead of temporal-only (frozen backbone) (ablation) | ViTAL-X · Avg-Temp | accuracy: +1.8 Avg-Temp | Improved |
| add margin temporal loss to contrastive loss (ablation) | ViTAL-X · XTE-Bench | accuracy: 69.4 with dual objective | Improved |
| use GPT-4o-mini instead of GPT-3.5-Turbo for counterfactual captions (ablation) | ViTAL-X · XTE-Bench | accuracy: 67.1 -> 69.4 | Improved |
| use open-source LLaMA-3-8B instead of GPT-4o-mini (ablation) | ViTAL-X · XTE-Bench | accuracy: 69.4 -> 66.3 | No clear effect |
| vary filtering thresholds (3x3 sweep) (ablation) | ViTAL-X · Avg-Temp | accuracy: +/-0.4 around default | No clear effect |
| scale to 7B multimodal LLMs without temporal supervision | Qwen2.5 / LLaVA-NeXT-Video-7B · XTE-Bench | accuracy: hovering near 56% | No clear effect |
Closest papers by meaning
- VideoComp: Advancing Fine-Grained Compositional and Temporal Alignment in Video-Text Models (CVPR 2025 · similarity 0.90)
- Dynamic Reflections: Probing Video Representations with Text Alignment (ICLR 2026 · similarity 0.89)
- TOPA: Extending Large Language Models for Video Understanding via Text-Only Pre-Alignment (NeurIPS 2024 · similarity 0.88)
- Inference-Time Text-to-Video Alignment with Diffusion Latent Beam Search (NeurIPS 2025 · similarity 0.87)
- ViTeX-Bench: Benchmarking High-Fidelity Video Scene Text Editing (NeurIPS 2026 · similarity 0.87)
- SteerVTE: Seamless Video Text Editing with Style and Glyph Control (NeurIPS 2026 · similarity 0.86)
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.
