The pretrained CNN that we will use is called Inception-v3. This deep CNN has been trained on the ImageNet image database (an academic benchmark for computer vision algorithms containing a vast library of labelled images covering a wide range of nouns) and can classify entire images into 1,000 classes found in everyday life, such as "pizza", "plastic bag", "red wine", "desk", "orange", and "basketball", to name just a few.
The Inception-v3 deep CNN was developed and trained by TensorFlow TM, an open source machine learning framework and software library for high-performance numerical computation originally developed within Google's AI organization.
To learn more about TensorFlow, Inception-v3, and ImageNet, please ...