The architecture of visual search

Deep neural network models such as ResNet, VGG16, and Inception can essentially be broken down into two components:

  • The first component identifies the low-level content of an image, such as feature vectors (edges).
  • The second component represents the high-level content of an image, such as the final image characteristics, which are an ensemble of various low-level contents. The following diagram illustrates a convolutional neural network that classifies seven classes:

The preceding diagram shows that the entire image classification neural network model can be divided into two: the convolution layer and the ...

Get Mastering Computer Vision with TensorFlow 2.x now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.