Mastering Computer Vision with TensorFlow 2.x

Constructing a model

In this section, we will import a model. The code for model construction is shown as follows (the explanation of each code snippet is just underneath the code):

model = Modelx(weights='imagenet', include_top=True,input_shape=(img_height, img_width, 3))Modelx = VGG16 or ResNet50 or InceptionV3

The model construction has three important parameters:

• weights is the pre-trained model we used on ImageNet images that were downloaded before.
• The include_top function indicates whether the final dense layer should be included or not. For class prediction for a pre-trained model, this is always True; however, in a later part of the chapter (the Developing a transfer learning model using TensorFlow section), we will learn that, ...