Mastering Computer Vision with TensorFlow 2.x

Landmark models

The following table describes models for age-gender and emotion recognition, facial landmark recognition, vehicle color and type recognition, and person attributes such as shirts, hats, and backpacks being worn. Note the input and output for each model listed in the table as this will need to be entered in the Python code for inference:

Here, the input is the same: there are different dimensions for different models, but the outputs are different. For landmarks, the output can be either 5 points or 35 points of contour in the face. For a person attribute, the output can be a binary result for each of the eight attributes. Vehicle ...