AI Paper Atlas

Research map / Video / Video segmentation and tracking

Temporal Cue Guided Video Highlight Detection With Low-Rank Audio-Visual Fusion

Qinghao Ye, Xiyue Shen, Yuan Gao, Zirui Wang, Qi Bi, Ping Li, Guang Yang

ICCV 2021 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 63 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 video-level supervision loss LV (Ls+Lvar -> full) (ablation)full model · YouTube HighlightsmAP: 0.6011 -> 0.6297Improved
add video-level supervision loss LV (Ls+Lvar -> full) (ablation)full model · TVSummAP: 0.7470 -> 0.7682Improved
add variation loss Lvar to score loss Ls (ablation)instance loss · YouTube HighlightsmAP: 0.5625 -> 0.6011Improved
add variation loss Lvar to score loss Ls (ablation)instance loss · TVSummAP: 0.6805 -> 0.7470Improved
add score loss Ls to variation loss Lvar (ablation)instance loss · YouTube HighlightsmAP: 0.5749 -> 0.6011Improved
add score loss Ls to variation loss Lvar (ablation)instance loss · TVSummAP: 0.7161 -> 0.7470Improved
use low-rank audio-visual tensor fusion instead of summation (ablation)feature fusion · YouTube HighlightsmAP: 0.5809 -> 0.6297Improved
use low-rank tensor fusion instead of concatenation (ablation)feature fusion · TVSummAP: 0.7187 -> 0.7682Improved
use low-rank fusion instead of compact bilinear pooling (saves 85% parameters) (ablation)feature fusion · YouTube HighlightsmAP: 0.6256 -> 0.6297Improved
use low-rank fusion on core tensor instead of factorized bilinear pooling (ablation)feature fusion · TVSummAP: 0.7511 -> 0.7682Improved
use attention-gated instance aggregation instead of Noisy-OR (ablation)video score modeling · YouTube HighlightsmAP: 0.5868 -> 0.6297Improved
use attention-gated instance aggregation instead of Noisy-OR (ablation)video score modeling · TVSummAP: 0.6838 -> 0.7682Improved

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.

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

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