AI Paper Atlas

Research map / Video / Video segmentation and tracking

Weakly Supervised Temporal Anomaly Segmentation With Dynamic Time Warping

Dongha Lee, Sehun Yu, Hyunjun Ju, Hwanjo Yu

ICCV 2021 · Video › Video segmentation and tracking · approach: Weakly supervised video anomaly detection · 18 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 DTW-based segmentation (ablation)WETAS (DiCNN) · SMDF1: 0.1265 -> 0.4358Improved
add alignment loss (with DTW segmentation) (ablation)WETAS (DiCNN) · SMDF1: 0.3384 -> 0.4358Improved
add alignment loss (without DTW segmentation) (ablation)WETAS (DiCNN) · SMDF1: 0.1225 -> 0.1265No clear effect
use DiCNN instead of LSTM (ablation)WETAS · SMDF1: 0.3685 -> 0.4358Improved
use global max pooling instead of average pooling (ablation)WETAS · SMDF1: 0.3046 -> 0.4358Improved
use corruption instead of random sampling for negative pseudo-label (ablation)WETAS · SMDF1: 0.4272 -> 0.4358Improved
use WETAS weak supervision instead of best DeepMILDeepMIL-4/8/16 · EMGF1: 0.4699 -> 0.5803Improved
use WETAS instead of best DeepMILDeepMIL-8 · SubwayF1: 0.6471 -> 0.7414Improved
use instance-level labels via weak supervision instead of unsupervised point-level detectorsOmniAnomaly++ · GHLF1: 0.1209 -> 0.2295Improved
train only on normal instances (semi-supervised ++ variant)OmniAnomaly · GHLF1: 0.0611 -> 0.1209Improved
train only on normal instances (semi-supervised ++ variant)OmniAnomaly · SMDF1-best: 0.1223 -> 0.1750Improved
train only on normal instances (semi-supervised ++ variant)Donut · SMDF1-best: 0.3206 -> 0.2875Hurt

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