AI Paper Atlas

Research map / Video / Video-language understanding

Flow Graph to Video Grounding for Weakly-Supervised Multi-step Localization

Nikita Dvornik, Isma Hadji, Hai Pham, Dhaivat Bhatt, Brais Martinez, Afsaneh Fazly, Allan D. Jepson

ECCV 2022 · Video › Video-language understanding · approach: Temporal video grounding · 21 citations (OpenAlex)

Read the paper

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
ground the flow graph with Graph2Vid instead of aligning a linear procedurestep localization at inference · CrossTaskAcc.: 22.3 -> 25.3Improved
ground the flow graph with Graph2Vid instead of aligning a linear procedurestep localization at inference · CrossTaskIoU: 15.1 -> 17.1Improved
use linear procedure order instead of unordered bag of stepsstep localization at inference · CrossTaskAcc.: 20.5 -> 22.3Improved
train with Graph2Vid loss instead of Drop-DTW on linear procedureMLP on pre-trained features · CrossTaskAcc.: 25.0 -> 26.3Improved
train with Graph2Vid loss instead of soft clustering on bag of stepsMLP on pre-trained features · CrossTaskIoU: 17.3 -> 19.1Improved
train with linear-procedure Drop-DTW versus no trainingMLP on pre-trained features · CrossTaskAcc.: 25.3 -> 25.0Hurt
train with Graph2Vid versus no trainingMLP on pre-trained features · CrossTaskIoU: 17.1 -> 19.1Improved
use learning-based parser graphs instead of rule-based parser graphs (ablation)Graph2Vid · CrossTaskAcc.: 24.3 -> 25.3Improved
use learning-based parser graphs instead of manual annotation (ablation)Graph2Vid · CrossTaskAcc.: 24.8 -> 25.3Improved
use manual graphs instead of rule-based parser graphs (ablation)Graph2Vid · CrossTaskIoU: 16.3 -> 16.8Improved
use tSort graph algorithm instead of brute-force enumeration of topological sortsflow graph grounding · CrossTaskgrounding time: 3.2 * 10^4 ms -> 57 msImproved
use ground-truth step order instead of flow graphstep localization at inference · CrossTaskAcc.: 25.3 -> 32.4Improved

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