AI Paper Atlas

Research map / Video / Video segmentation and tracking

BEVT: BERT Pretraining of Video Transformers

Rui Wang, Dongdong Chen, Zuxuan Wu, Yinpeng Chen, Xiyang Dai, Mengchen Liu, Yu-Gang Jiang, Luowei Zhou, Lu Yuan

CVPR 2022 · Video › Video segmentation and tracking · approach: Self-supervised video representation learning · 214 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 video-stream masked modeling to image-stream pretrainingBEVT-I · SSv2Top-1 accuracy: 69.2 -> 70.6Improved
Add video-stream masked modeling to image-stream pretrainingBEVT-I · Diving-48Top-1 accuracy: 81.2 -> 86.7Improved
Add video-stream masked modeling to image-stream pretrainingBEVT-I · K400Top-1 accuracy: 80.5 -> 80.6No clear effect
Initialize the video stream with image-stream pretrained weights (ablation)BEVT-V · SSv2Top-1 accuracy: 67.1 -> 70.0Improved
Initialize the video stream with image-stream pretrained weights (ablation)BEVT-V · Diving-48Top-1 accuracy: 83.7 -> 85.2Improved
Initialize the video stream with image-stream pretrained weights (ablation)BEVT-V · K400Top-1 accuracy: 76.2 -> 79.6Improved
Initialize joint BEVT with image-stream pretrained weights (ablation)BEVT · SSv2Top-1 accuracy: 67.9 -> 70.6Improved
Initialize joint BEVT with image-stream pretrained weights (ablation)BEVT · Diving-48Top-1 accuracy: 85.1 -> 86.7Improved
Initialize joint BEVT with image-stream pretrained weights (ablation)BEVT · K400Top-1 accuracy: 78.5 -> 80.6Improved
Use ImageNet images rather than K400 frames for the image stream (ablation)BEVT joint pretraining · SSv2Top-1 accuracy: 70.3 -> 70.6Improved
Use ImageNet images rather than K400 frames for the image stream (ablation)BEVT joint pretraining · Diving-48Top-1 accuracy: 86.6 -> 86.7Improved
Use ImageNet images rather than K400 frames for the image stream (ablation)BEVT joint pretraining · K400Top-1 accuracy: 80.5 -> 80.6Improved

Later papers that compared with this method

2 later accepted papers report BEVT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.

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