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 | 31 |
- simclrv2
- Pseudo Label
- 백준 알고리즘
- mocov3
- SSL
- dann paper
- shrinkmatch
- BYOL
- CoMatch
- 컴퓨터구조
- cifar100-c
- semi supervised learnin 가정
- ConMatch
- WGAN
- 최린컴퓨터구조
- conjugate pseudo label paper
- UnderstandingDeepLearning
- CycleGAN
- shrinkmatch paper
- Entropy Minimization
- dcgan
- tent paper
- mme paper
- Meta Pseudo Labels
- Pix2Pix
- GAN
- CGAN
- adamatch paper
- 딥러닝손실함수
- remixmatch paper
- Today
- Total
목록Pix2Pix (2)
Hello Data

지난번 논문 리뷰에 이은 코드리뷰이다. 생성자와 판별자, 손실함수 위주로 진행해보겠습니다. 생성자 정의 class UNetDown(nn.Module): #UNet class 정의하기 def __init__(self, in_channels, out_channels, normalize = True, dropout = 0.0): super(UNetDown, self).__init__() layers = [nn.Conv2d(in_channels, out_channels,4, stride = 2, padding = 1, bias = False)] if normalize: layers.append(nn.InstanceNorm2d(out_channels)) layers.append(nn.LeakyReLU(0.2)) ..

지난 번 U-Net에 이어서 U-Net구조를 활용한 Pix2Pix 논문 리뷰입니다. 처음부터 천천히 읽어나가보겠습니다. Introduction 처음부터 translation에 대하여 정의하는데요, translation an input image into a corresponding output image로 정의할 수 있지만 저자들은 translating one possible representation of a scene into another 으로 정의합니다. 요약해보자면 optimal 한 G가 있다한다면 A이미지에 대해 B로도 translating 가능하고 C로도 translating될 수 있다는 가능성이 있다는 걸 말하는 거 같습니다. 본인들이 만든 framework에 대해서 CNN을 사용하는데 이..