AI Paper Atlas

Research map / Video / Video segmentation and tracking

MGMAE: Motion Guided Masking for Video Masked Autoencoding

Bingkun Huang, Zhiyu Zhao, Guozhen Zhang, Yu Qiao, Limin Wang

ICCV 2023 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 53 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
replace tube masking with motion guided maskingVideoMAE ViT-B · Something-Something V2 (800e)Acc@1: 69.6 -> 71.0Improved
replace tube masking with motion guided masking, longer pre-trainingVideoMAE ViT-B · Something-Something V2 (2400e)Acc@1: 70.8 -> 72.3Improved
replace tube masking with motion guided maskingVideoMAE ViT-B · Kinetics-400 (1600e)Acc@1: 81.5 -> 81.8Improved
replace tube masking with motion guided maskingVideoMAE ViT-B · Kinetics-400 (800e)Acc@1: 80.0 -> 81.2Improved
use middle frame as base frame instead of first or random frame (ablation)MGMAE · SSV2 ablationAcc@1: 70.5 -> 71.0Improved
use backward warping instead of forward warping (ablation)MGMAE · SSV2 ablationAcc@1: 70.5 -> 71.0Improved
frame-level top-k sampling instead of clip-level (ablation)MGMAE · SSV2 ablationAcc@1: 70.7 -> 71.0Improved
mixed Gaussian initial mask instead of token or pixel random (ablation)MGMAE · SSV2 ablationAcc@1: 70.9 -> 71.0Improved
tube hole filling instead of random, invisible, visible or previous-map filling (ablation)MGMAE · SSV2 ablationAcc@1: 70.6 -> 71.0Improved
use TVL1 offline flow instead of online RAFT (ablation)MGMAE · SSV2 ablationAcc@1: 71.2 vs 71.0No clear effect
reduce RAFT iterations from 12 to 6 (ablation)MGMAE · SSV2 ablationAcc@1: 71.0 -> 71.3 (56 h -> 43 h)Improved
add motion guided masking (costs extra pre-training time)VideoMAE · SSV2 ablationpre-training time: 32 h -> 43 hHurt

Later papers that compared with this method

3 later accepted papers report MGMAE as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

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