AI Paper Atlas

Research map / Video / Video segmentation and tracking

Dual-Path Adaptation From Image to Video Transformers

Jungin Park, Jiyoung Lee, Kwanghoon Sohn

CVPR 2023 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 49 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 the temporal adaptation path (ablation)DualPath · Something-Something-v2R@1: 53.7 -> 69.3Improved
add the spatial adaptation path (ablation)DualPath · Something-Something-v2R@1: 55.1 -> 69.3Improved
increase spatial frames from T_S=8 to T_S=12 (ablation)DualPath · Something-Something-v2R@1: 69.3 -> 69.6Improved
increase spatial frames from T_S=8 to T_S=16 (ablation)DualPath · Something-Something-v2R@1: 69.3 -> 69.8Improved
set grid scaling factors to w=h=2 instead of w=h=4 (ablation)DualPath · Something-Something-v2R@1: 71.8 -> 66.4Hurt
set grid scaling factors to w=h=8 instead of w=h=4 (ablation)DualPath · Something-Something-v2R@1: 71.8 -> 61.5Hurt
increase grid resolution from 224×224 to 448×448 (ablation)DualPath · Something-Something-v2R@1: 69.3 -> 70.5Improved
increase grid resolution from 224×224 to 896×896 (ablation)DualPath · Something-Something-v2R@1: 69.3 -> 71.6Improved
use 48 frames instead of 16 at 224×224 grid resolution (ablation)DualPath · Something-Something-v2R@1: 69.3 -> 71.2Improved
use DualPath with ViT-L/14 rather than X-CLIP-L/14Kinetics-400 · Kinetics-400R@1: 87.1 -> 87.7Improved
use DualPath rather than ST-Adapter with ViT-B/16ViT-B/16 · Something-Something-v2R@1: 69.5 -> 70.3Improved
use DualPath rather than VideoPromptViT-B/16 · HMDB51accuracy: 66.4 -> 75.6Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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