AI Paper Atlas

Research map / Video / Video-language understanding

Streamlined Dense Video Captioning

Jonghwan Mun, Linjie Yang, Zhou Ren, Ning Xu, Bohyung Han

CVPR 2019 · Video › Video-language understanding · approach: Dense video captioning · 159 citations (OpenAlex)

Check 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 changedWhereMeasuredEffect
Select an ordered event sequence with ESGN instead of using all candidate proposals (ablation)Dense video captioning · ActivityNet Captions validationMETEOR: 4.58 -> 6.73Improved
Select an ordered event sequence with ESGN instead of using all candidate proposals (ablation)Event proposal selection · ActivityNet Captions validationNumber of proposals: 77.99 -> 2.85Improved
Add sequential captioning with episode-level context to independent captions (ablation)ESGN captioning · ActivityNet Captions validationMETEOR: 6.73 -> 6.92Improved
Add reinforcement learning to sequential captioning (ablation)ESGN-SCN · ActivityNet Captions validationMETEOR: 6.92 -> 8.82Improved
Use event- and episode-level rewards together rather than event-level reward alone (ablation)SDVC reinforcement learning · ActivityNet Captions datasetMETEOR: 8.73 -> 8.82Improved
Add event-level reward to episode-level reward (ablation)SDVC reinforcement learning · ActivityNet Captions datasetMETEOR: 8.29 -> 8.82Improved
Use ESGN rather than MFT for event sequence selectionEvent proposal selection · ActivityNet Captions validationRecall (average): 24.31 -> 55.58Improved
Use SDVC rather than MFT for dense captioning with learned proposalsDense video captioning · ActivityNet Captions validationMETEOR: 7.08 -> 8.82Improved
Use SDVC rather than TDA-CG with learned proposalsDense video captioning · ActivityNet Captions validationBLEU@4: 1.31 -> 0.93Hurt
Use SDVC rather than Masked Transformer with GT proposalsDense video captioning · ActivityNet Captions validationCIDEr: 47.71 -> 43.48Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy