I will share with you my journey in building Qrio, an AI bot that can interact and speak to our toddler, recognises toys he carry and plays YouTubeKids video about the toy. Eg. Airplane, panda, etc. A God send for a working home parents during COVID-19 with an active toddler who is super curious! I will talk about how I train an Object Detection AI model to recognises toys and person followed by how I optimise the model on TensorRT. Then I will discuss how I build the text to speech component with Amazon Polly. I will also discuss how I build the Qrio body and animation system using a game engine and a skeleton animation technique. I will then continue to discuss how I programatically play a YouTube video using Selenium Automation tools. Finally I will show you how I connect everything together and getting it to run on an NVIDIA Jetson Nano and Raspberry camera.
Get notified about new features and conference additions.