Research map / Video / Video segmentation and tracking
MoVie: Multimodal Video Compression with Text Guidance
Jiaqi Hu, Haoji Hu, Heming Sun, Lianrui Mu
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 text guidance in the encoder only (ablation) | Vie · UVG autoencoder comparison (no entropy coding) | FID: 0.0062 -> 0.0041 | Improved |
| add text guidance in the encoder only (ablation) | Vie · UVG autoencoder comparison (no entropy coding) | LPIPS: 0.0018 -> 0.0009 | Improved |
| add text guidance in the encoder only (ablation) | Vie · UVG autoencoder comparison (no entropy coding) | PSNR: 46.78 -> 46.49 | Hurt |
| add text to the decoder as well as the encoder (ablation) | MoVie · UVG autoencoder comparison (no entropy coding) | FID: 0.0041 -> 0.0057 | Hurt |
| add text to the decoder only instead of encoder (ablation) | MoVie · UVG autoencoder comparison (no entropy coding) | PSNR: 46.49 -> 45.08 | Hurt |
| use frame-specific two-stage text fusion instead of Global Text (ablation) | MoVie · UVG autoencoder comparison (no entropy coding) | FID: 1.6403 -> 0.0041 | Improved |
| use two-stage fusion instead of Copy Text (ablation) | MoVie · UVG autoencoder comparison (no entropy coding) | FID: 0.0212 -> 0.0041 | Improved |
| use informative captions instead of constant fixed caption (ablation) | MoVie · UVG autoencoder comparison (no entropy coding) | FID: 0.0593 -> 0.0041 | Improved |
| use accurate captions instead of irrelevant captions (ablation) | MoVie · UVG text perturbation robustness | FID: 15.129 -> 6.731 | Improved |
| use accurate captions instead of no caption (ablation) | MoVie · UVG text perturbation robustness | FID: 14.694 -> 6.731 | Improved |
| add S-C entropy model to VCT baseline (ablation) | VCT baseline · UVG progressive ablation | BD-Rate LPIPS: 0.00% -> -45.39% | Improved |
| add VideoTCM autoencoder (ablation) | VCT + S-C entropy · UVG progressive ablation | BD-Rate PSNR: -5.70% -> -19.51% | Improved |
Closest papers by meaning
- TTVidT: Decoupling the Temporal Axis for Efficient Motion-Centric Video Pretraining (NeurIPS 2026 · similarity 0.87)
- When MLLMs Meets Compression Distortion: A Coding Paradigm Tailored to MLLMs (ICLR 2026 · similarity 0.87)
- CoPE-VideoLM: Leveraging Codec Primitives For Efficient Video Language Modelling (NeurIPS 2026 · similarity 0.87)
- AdaCodec: A Predictive Visual Code for Video MLLMs (NeurIPS 2026 · similarity 0.87)
- Snap Video: Scaled Spatiotemporal Transformers for Text-to-Video Synthesis (CVPR 2024 · similarity 0.86)
- Encapsulated Composition of Text-to-Image and Text-to-Video Models for High-Quality Video Synthesis (CVPR 2025 · 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.
