Hands-On Python Deep Learning for the Web

Expecting the AI part of the website to be real time

AI is computationally expensive and needless to say, this is undesirable for a website that aims to serve its clients in the quickest time possible. While smaller models or using browser AI (such as TensorFlow.js or other libraries) can provide the experience of real-time AI responses, even they suffer issues where the client is in a slow network area or using a low-end device. So, both the methods of in-browser AI models or lightweight AI models replying near instantaneously are subject to device configuration and network bandwidth. Hence, the backend of the website, which is supposed to make quick responses to the client, should ideally be separated from the part that handles the AI model ...