Research map / Video / Video-language understanding
Vamos: Versatile Action Models for Video Understanding
Shijie Wang, Qi Zhao, Minh Quan Do, Nakul Agarwal, Kwonjoon Lee, Chen Sun
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 caption text instead of CLIP visual embeddings (ablation) | Vamos · NeXT-QA val | accuracy (All): 69.6 -> 75.0 | Improved |
| add visual embeddings to caption text (ablation) | Vamos · NeXT-QA val | accuracy (All): 75.0 -> 74.5 | No clear effect |
| add visual embeddings to caption text (ablation) | Vamos · IntentQA test | accuracy (All): 73.2 -> 71.7 | Hurt |
| add visual embeddings to action text on Ego4D LTA (ablation) | Vamos · Ego4D LTA test | action edit distance: 0.878 -> 0.868 | Improved |
| use token bottleneck selecting 40 of 644 tokens (ablation) | Vamos · NeXT-QA val | accuracy (All): 77.3 -> 69.6 | Mixed |
| increase selected tokens from 20 to 40 (ablation) | Token bottleneck model · NeXT-QA val | accuracy (All): 67.4 -> 69.6 | Improved |
| use LLaVA-1.5 captions instead of BLIP-2 (ablation) | Vamos · NeXT-QA val | accuracy (All): 72.4 -> 75.0 | Improved |
| scale caption model from LLaVA 7B to 13B (ablation) | Vamos · NeXT-QA val | accuracy (All): 75.1 -> 75.0 | No clear effect |
| increase captioned frames from 1 to 6 (ablation) | Vamos · NeXT-QA val | accuracy (All): 69.8 -> 75.0 | Improved |
| increase captioned frames from 1 to 12 (ablation) | Vamos (GPT-3.5) · EgoSchema zero-shot | accuracy: 37.83% -> 41.24% | Improved |
| upgrade LLM from LLaMA1-7B to LLaMA3-8B (ablation) | Vamos · NeXT-QA val | accuracy (All): 75.0 -> 77.3 | Improved |
| use captions instead of action labels as text representation (ablation) | Vamos (GPT-4) · EgoSchema zero-shot | accuracy: 38.12% -> 48.26% | Improved |
Later papers that compared with this method
1 later accepted paper reports Vamos as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Streaming Long Video Understanding with Large Language Models (NeurIPS 2024 · better on 0 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- AntGPT — AntGPT: Can Large Language Models Help Long-term Action Anticipation from Videos? (ICLR 2024) · better on 1 of 1 shared result
- Palm — PALM: Predicting Actions through Language Models (ECCV 2024) · 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
- Not Another Text Benchmark: Putting the “Visual" Back in Visual Question Answering for Large Video Models (NeurIPS 2026 · similarity 0.92)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.91)
- Revisiting the "Video" in Video-Language Understanding (CVPR 2022 · similarity 0.91)
- Leveraging Textual Memory and Key Frame Reasoning for Full Video Understanding Using Off-the-Shelf LLMs and VLMs (Student Abstract) (AAAI 2025 · similarity 0.90)
- LongVLM: Efficient Long Video Understanding via Large Language Models (ECCV 2024 · similarity 0.90)
- Language-aware Visual Semantic Distillation for Video Question Answering (CVPR 2024 · similarity 0.90)
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.
