Mastering Computer Vision with TensorFlow 2.x

Converting the pre-trained model for inference

This section describes the steps involved in converting a custom TensorFlow model created using the TensorFlow Keras object classification model we developed in Chapter 6, Visual Search Using Transfer Learning, or using a model created using the TensorFlow Object Detection API, as we did in the previous chapter. The steps described in the previous section will work if you already plan on using a pre-trained optimized model from Intel Open Source Technology Center. In the next section, we'll describe how to perform conversion using two types of TensorFlow models.