Upstage AI Lab 2기
2024년 3월 8일 (금) Day_060
온라인 강의
nn.ModuleList 사용해서 VGG 구현해보기
Simonyan, Karen, and Andrew Zisserman.
"Very deep convolutional networks for large-scale image recognition." arXiv preprint arXiv:1409.1556 (2014).
large-scale image recognition setting
input : 224 × 224 RGB image
For instance, the best-performing submissions to the ILSVRC2013
(Zeiler & Fergus, 2013; Sermanet et al., 2014) utilised smaller receptive window size and
smaller stride of the first convolutional layer.
In this paper, we address another important aspect of ConvNet architecture
design – its depth.
https://www.robots.ox.ac.uk/~vgg/research/very_deep/
The only preprocessing
we do is subtracting the mean RGB value, computed on the training set, from each pixel.
In one of the configurations we also utilise 1 × 1 convolution filters, which can be seen as
a linear transformation of the input channels (followed by non-linearity).
The convolution stride is
fixed to 1 pixel; the spatial padding of conv. layer input is such that the spatial resolution is preserved
after convolution, i.e. the padding is 1 pixel for 3 × 3 conv. layers.
2.2 CONFIGURATIONS 부터 다시 읽기
VGG 에 cifar 10을 넣는 시도를 하는게 맞는지,
그냥 resize 해버리면 되는지
'Upstage AI Lab 2기' 카테고리의 다른 글
Upstage AI Lab 2기 [Day062] (0) | 2024.03.12 |
---|---|
Upstage AI Lab 2기 [Day061] (0) | 2024.03.11 |
Upstage AI Lab 2기 [Day060] (0) | 2024.03.08 |
Upstage AI Lab 2기 [Day060] 수강생 TODO #02 - Pytorch Data Loading - 수정작업 #2 (4) | 2024.03.08 |
Upstage AI Lab 2기 [Day059] 수강생 TODO #02 - Pytorch Data Loading - 수정작업 (0) | 2024.03.07 |