둘러보기로 가기 검색하러 가기



  1. VGGNet is a Convolutional Neural Network architecture proposed by Karen Simonyan and Andrew Zisserman from the University of Oxford in 2014.[1]
  2. Another variation of VGGNet has 19 weight layers consisting of 16 convolutional layers with 3 fully connected layers and same 5 pooling layers.[1]
  3. In both variation of VGGNet there consists of two Fully Connected layers with 4096 channels each which is followed by another fully connected layer with 1000 channels to predict 1000 labels.[1]
  4. VGGNet is a neural network that performed very well in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) in 2014.[2]
  5. VGGNet apparently took 2-3 weeks to train on a computer with four NVIDIA Titan Black GPUs.[2]
  6. Let’s say you want to train a network such as VGGNet to recognize faces of celebrities.[2]
  7. Since the VGGNet we’re using was trained on ImageNet, it’s really good at distinguishing between different breeds of dogs, different types of fish, and so on.[2]
  8. VGGNet is invented by Visual Geometry Group (by Oxford University).[3]
  9. The reason to understand VGGNet is that many modern image classification models are built on top of this architecture.[3]
  10. It achieves better accuracy than VGGNet and GoogLeNet while being computationally more efficient than VGGNet.[4]
  11. The architecture is similar to the VGGNet consisting mostly of 3X3 filters.[4]
  12. From the VGGNet, shortcut connection as described above is inserted to form a residual network.[4]
  13. VGGNet was a competitor in the ImageNet ILSVRC-2014 image classification competition and scored second place.[5]
  14. The runner-up in ILSVRC 2014 was the network from Karen Simonyan and Andrew Zisserman that became known as the VGGNet.[6]
  15. A downside of the VGGNet is that it is more expensive to evaluate and uses a lot more memory and parameters (140M).[6]
  16. Lets break down the VGGNet in more detail as a case study.[6]
  17. The whole VGGNet is composed of CONV layers that perform 3x3 convolutions with stride 1 and pad 1, and of POOL layers that perform 2x2 max pooling with stride 2 (and no padding).[6]
  18. They compare HybridNet, with VGGNet and CBP-CNN, for 292, 100, and 200 sub-classes of VegFru Dataset.[7]