article2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)Jun 1, 2022Closed access

Masked Feature Prediction for Self-Supervised Visual Pre-Training

CWChen WeiHFHaoqi FanSXSaining XieCWChao-Yuan WuAYAlan Yuille

Johns Hopkins University · Meta (Israel)

Indexed incrossref

Abstract

We present Masked Feature Prediction (MaskFeat) for self-supervised pre-training of video models. Our approach first randomly masks out a portion of the input sequence and then predicts the feature of the masked regions. We study five different types of features and find Histograms of Oriented Gradients (HOG), a hand-crafted feature descriptor, works particularly well in terms of both performance and efficiency. We observe that the local contrast normalization in HOG is essential for good results, which is in line with earlier work using HOG for visual recognition. Our approach can learn abundant visual knowledge and drive large-scale Transformer based models. Without using extra model weights or supervision,…

Citation impact

526
total citations
FWCI
29.74
Percentile
100%
References
118
Citations per year

Authors

6

Topics & keywords

Topics
Keywords
  • Artificial intelligence
  • Computer science
  • Pattern recognition (psychology)
  • Normalization (sociology)
  • Histogram
  • Feature (linguistics)
  • Feature extraction
  • Computer vision
No related works found for this paper.

Funding