Research map / Video / Video generation
DiffusionRet: Generative Text-Video Retrieval with Diffusion Model
Peng Jin, Hao Li, Zesen Cheng, Kehan Li, Xiangyang Ji, Chang Liu, Li Yuan, Jie Chen
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 |
|---|---|---|---|
| use KL divergence instead of MSE as generation loss (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 46.9 -> 49.0 | Improved |
| use DDIM instead of DDPM sampling (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 48.9 -> 49.0 | No clear effect |
| use cosine instead of linear beta schedule (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 48.7 -> 49.0 | Improved |
| add discriminative training to generative training (hybrid) (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 40.2 -> 49.0 | Improved |
| add generative training to discriminative training (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 46.8 -> 49.0 | Improved |
| use 50 diffusion steps instead of 1000 for training (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 48.6 -> 49.0 | Improved |
| scale of beta 1.0 versus other scales in 0.1 to 2.0 (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 48.3 -> 49.0 | Improved |
| increase inference diffusion steps from 1 to 50 (ablation) | DiffusionRet · MSRVTT 1k-A text-to-video (gallery 1000) | R@1: 23.4 -> 49.0 | Improved |
| use generative DiffusionRet instead of discriminative baselines for out-domain retrieval | text-video retrieval · MSRVTT->DiDeMo | R@1: 31.8 -> 33.2 | Improved |
| add QB-Norm to DiffusionRet | DiffusionRet · ActivityNet Captions text-to-video | R@1: 45.8 -> 48.1 | Improved |
| add QB-Norm to DiffusionRet | DiffusionRet · MSRVTT 1k-A text-to-video | R@1: 49.0 -> 48.9 | No clear effect |
Later papers that compared with this method
7 later accepted papers report DiffusionRet as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- From Script to Shot: A Benchmark for Grounding Screenplays in Movies (ECCV 2026 · better on 2 of 2 shared results)
- Distribution-Alignment Bridge for Uncertainty-Aware Text-to-Video Retrieval (ECCV 2026 · better on 2 of 2 shared results)
- Learning Dynamic Similarity by Bidirectional Hierarchical Sliding Semantic Probe for Efficient Text Video Retrieval (AAAI 2025 · better on 2 of 3 shared results)
- Rebalancing Contrastive Alignment with Bottlenecked Semantic Increments in Text-Video Retrieval (NeurIPS 2025 · better on 2 of 2 shared results)
- MUSE: Mamba Is Efficient Multi-scale Learner for Text-video Retrieval (AAAI 2025 · better on 1 of 1 shared result)
- Text Is MASS: Modeling as Stochastic Embedding for Text-Video Retrieval (CVPR 2024 · better on 1 of 1 shared result)
- KDProR: A Knowledge-Decoupling Probabilistic Framework for Video-Text Retrieval (ECCV 2024 · better on 1 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TS2-Net — TS2-Net: Token Shift and Selection Transformer for Text-Video Retrieval (ECCV 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Diffusion-Inspired Truncated Sampler for Text-Video Retrieval (NeurIPS 2024 · similarity 0.91)
- Generative Video Diffusion for Unseen Novel Semantic Video Moment Retrieval (AAAI 2025 · similarity 0.88)
- FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling (ICLR 2024 · similarity 0.88)
- VidProM: A Million-scale Real Prompt-Gallery Dataset for Text-to-Video Diffusion Models (NeurIPS 2024 · similarity 0.88)
- Divot: Diffusion Powers Video Tokenizer for Comprehension and Generation (CVPR 2025 · similarity 0.87)
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · 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.
