AI Paper Atlas

Research map / Video / Video segmentation and tracking

Stand-Alone Inter-Frame Attention in Video Models

Fuchen Long, Zhaofan Qiu, Yingwei Pan, Ting Yao, Jiebo Luo, Tao Mei

CVPR 2022 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 72 citations (OpenAlex)

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 SIFA temporal modeling to ResNet-50ResNet-50 · Kinetics-400 validationTop-1 accuracy: 72.0 -> 77.4Improved
add inter-frame attention over local regions (ablation)SIFA-C · Kinetics-400 validationTop-1 accuracy: 73.3 -> 74.6Improved
add deformable feature resampling (ablation)SIFA-R · Kinetics-400 validationTop-1 accuracy: 74.6 -> 75.4Improved
aggregate deformable neighbors from both previous and next frames (ablation)SIFA · Kinetics-400 validationTop-1 accuracy: 0.1% in top-1 accuracyImproved
predict deformable offsets from next-frame features (ablation)SIFA-R · Kinetics-400 validationTop-1 accuracy: 74.6 -> 74.7Improved
predict offsets using 3D convolution over the clip (ablation)SIFA-R · Kinetics-400 validationTop-1 accuracy: 74.7 -> 74.8Improved
predict offsets from temporal differences (ablation)SIFA-R · Kinetics-400 validationTop-1 accuracy: 74.8 -> 75.0Improved
use the motion saliency map to predict offsets (ablation)SIFA · Kinetics-400 validationTop-1 accuracy: 75.0 -> 75.4Improved
expand local attention from 1×1 to 3×3 (ablation)SIFA · Kinetics-400 validationTop-1 accuracy: 73.4 -> 75.4Improved
increase local region size beyond 3×3 (ablation)SIFA · Kinetics-400 validationTop-1 accuracy: 75.4; 75.4; 75.5No clear effect
add SIFA blocks in res3 and res4 (ablation)SIFA-Net with a block in res5 · Kinetics-400 validationTop-1 accuracy: 75.4 -> 77.4Improved
add a SIFA block in res2 (ablation)SIFA-Net with SIFA in res3–res5 · Kinetics-400 validationTop-1 accuracy: 77.4 -> 77.4No clear effect

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