y-unique☼

  • 홈
  • 태그
  • 방명록

vit 1

[논문리뷰] An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

진짜진짜 오랜만에 돌아온 논문리뷰 ViT를 가지고 왔습니당 정확하지 않으며 제 생각이 많이 들어간 글이라는 점.. 인지하시고~설 연휴 직전이걸랑요새복많입니다. Abstractnlp에서 transformer가 자리를 잡는 동안 cv에선 한계가 있었음cnn 구조를 가져가면서 attention을 적용하거나 전체 구조는 유지해야만 했음저자는 pure transformer에 직접적으로 Image patch를 붙여 image classification을 수행하도록 함많은 양의 데이터로 pre-trained 되었거나 imageNet 등의 인식 벤치마크로 trnasfered 한 경우 sota 달성대체로 fewer computational resourcesIntroductionCNN architecture with se..

논문 리뷰 2025.01.24
이전
1
다음
더보기
프로필사진

y-unique☼

🌀🌀

  • 분류 전체보기 (27)
    • 알고리즘 (1)
    • EECS 498-007 (13)
    • 논문 리뷰 (4)
    • 모각코 (9)

Tag

CNN Architectures, visition, Sementic Segmentation, EECS 498-007, RoI Aling, Contrastive learning, 알고리즘 #hash, Inductive Bias, data preprocessing, zero-shot, gram matrix, Training Neural Networks #Activation function, upsmapling, EECS 498-007 #10강 #, weight initialization, Mask R-CNN, style loss, content loss, EECS 498-007 3강 #Linear Classifier #SVM #Cross-Entropy #L1 #L2, EECS 498-007 #Optimization,

최근글과 인기글

  • 최근글
  • 인기글

최근댓글

공지사항

페이스북 트위터 플러그인

  • Facebook
  • Twitter

Archives

Calendar

«   2025/06   »
일 월 화 수 목 금 토
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30

방문자수Total

  • Today :
  • Yesterday :

Copyright © Kakao Corp. All rights reserved.

티스토리툴바