long8v / PTIR

Paper Today I Read
19 stars 0 forks source link

[63] Masked Autoencoders Are Scalable Vision Learners #69

Open long8v opened 2 years ago

long8v commented 2 years ago
image

paper

TL;DR

Details

Architecture

image

Result

image

target을 noramalize 한게 더 잘됐음(전체 패치의 평균과 분산으로 normalize)

Comparison with other SSL methods

image

mask ratio 높여도 잘된다.

image

얼룩말 한마리 된거 신기 ㅋㅋ

image

근데 75%가 잘되긴 함