AI Paper Atlas

Research map / Video / Action recognition

Motion-Augmented Self-Training for Video Recognition at Smaller Scale

Kirill Gavrilyuk, Mihir Jain, Ilia Karmanov, Cees G. M. Snoek

ICCV 2021 · Video › Action recognition · approach: Contrastive video representation learning · 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
use motion (optical flow) representation instead of appearance for the pseudo-label generator (ablation)M_MPLG · UCF101 split 1top-1 accuracy: 59.1 -> 82.6Improved
add multi-clip training (R=3) instead of single clip (ablation)M_MPLG motion · UCF101 split 1top-1 accuracy: 78.2 -> 82.6Improved
use 3 clips rather than longer 32-frame input (ablation)M_MPLG motion · UCF101 split 1top-1 accuracy: 80.8 -> 82.6Improved
add multi-clip training to appearance representation (ablation)M_MPLG appearance · UCF101 split 1top-1 accuracy: 58.9 -> 59.1No clear effect
add clip-order prediction term (lambda<1) to motion generator (ablation)M_MPLG motion · UCF101 split 1top-1 accuracyNo clear effect
use clip-level instead of video-level pseudo-labels (ablation)MotionFit self-training · UCF101 split 1top-1 accuracy: 76.5 -> 80.3Improved
use clip-level instead of segment-level pseudo-labels (ablation)MotionFit self-training · UCF101 split 1top-1 accuracy: 79.0 -> 80.3Improved
add next/previous clip pseudo-label prediction (ablation)MotionFit self-training · UCF101 split 1top-1 accuracy: +0.4%Improved
increase number of clusters from 1000 to 1600 (ablation)MotionFit (Kinetics-val) · UCF101 split 1top-1 accuracy: 79.7 -> 74.2Hurt
increase number of clusters from 128 to 1000 (ablation)MotionFit (Kinetics-train) · UCF101 split 1top-1 accuracy: 85.2 -> 86.5Improved
fine-tune with 32 frames instead of 16MotionFit · UCF101top-1 accuracy: 87.4 -> 88.9Improved
fine-tune with 32 frames instead of 16MotionFit · HMDB51top-1 accuracy: 56.4 -> 61.4Improved

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