AI Paper Atlas

Research map / Video / Video-language understanding

Video Re-localization

Yang Feng, Lin Ma, Wei Liu, Tong Zhang, Jiebo Luo

ECCV 2018 · Video › Video-language understanding · approach: Dense video captioning · 71 citations (OpenAlex)

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 changedWhereMeasuredEffect
add cross gating to the base matching model (ablation)base model · Video re-localization ablationmAP@1 average: 23.7 -> 24.3Improved
add factorized bilinear matching to the base model (ablation)base model · Video re-localization ablationmAP@1 average: 23.7 -> 25.0Improved
add bilinear matching on top of cross gating (ablation)cross gated model · Video re-localization ablationmAP@1 average: 24.3 -> 25.7Improved
add cross gating on top of bilinear matching (ablation)bilinear model · Video re-localization ablationmAP@1 average: 25.0 -> 25.7Improved
add cross gating to the base model at tIoU 0.5 (ablation)base model · Video re-localization ablationmAP@1 tIoU 0.5: 40.8 -> 40.5No clear effect
add cross gating to the base model at tIoU 0.9 (ablation)base model · Video re-localization ablationmAP@1 tIoU 0.9: 6.4 -> 6.1No clear effect
add bilinear matching to the base model at tIoU 0.8 (ablation)base model · Video re-localization ablationmAP@1 tIoU 0.8: 15.9 -> 15.4No clear effect
use cross gated bilinear matching instead of the SST proposal baselineSST · Video re-localization (ActivityNet-based, unseen test classes)mAP@1 average: 17.1 -> 25.7Improved
use cross gated bilinear matching instead of the video-level baselinevideo-level baseline · Video re-localization (ActivityNet-based, unseen test classes)mAP@1 average: 12.4 -> 25.7Improved
use learned video-level encoding instead of training-free frame-level matchingframe-level baseline · Video re-localization (ActivityNet-based, unseen test classes)mAP@1 average: 9.9 -> 12.4Improved

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