From the course: Innovating with AI Design Using Autonomous Vehicles

Demo: Designing for computer vision

From the course: Innovating with AI Design Using Autonomous Vehicles

Start my 1-month free trial Buy for my team

Demo: Designing for computer vision

“

- [Instructor] I want to show you a distracted driver data set that is captured by cameras inside the car and used by computer vision. Both of these are my own images that I built out for a research project I did. So one of them here shows the driver with their hands off the wheel, and they're scratching their face. This one, the hands is off the wheel again, but I would imagine the person has long hair, and the hair is flying as soon as they turn on the fan. So the important thing when you look at data for computer vision is that it's images. There is spatial dimension to the images in how the AI is looking at it. But as a designer, when you apply AI as framework, the AI is going to see the image. The user likely will not see the image. In this case, the user will get the design experience of what you present to them. So let us take a step back and think about the AI X framework. We start with the user. In this case, this is the driver, and we think about the use case. Are we helping the driver with driver safety? It feels like that's the very obvious use case. The third thing is there is an AI persona or AI character. So what's the value of that AI? What do you want the AI to do? It has to go back to understanding the user, thinking about them with empathy. So in this case, the user is scratching their face, taking their hands off the wheel, and if you want to ensure their safety and remind them to put their hands back on the wheel, how would you create that driver experience? In some cars, they create a vibration. Sometimes there might be a voice saying that, or it might just be the car taking over the control in case the driver is swerving off the lane. So those are all part of the experience of the user, but coming directly as AI X design based on the AI's input saying this driver is distracted. In this picture here, the user is turning on the fan and is actually not truly distracted, but their hair is flying, and the computer vision camera, the AI, might conclude that they're distracted. So in this case, how would you create that experience? And I would say this person is more distracted than this person who's turning on the fan. And if you're thinking about safety as the use case, this person is more safe than this person. So how would you create the experience? You can use the same vibration in the wheels, in the steering wheel or an announcement by voice or taking over the steering wheel control. But go back to thinking about the user with empathy. What is the headspace of the user? Maybe it is very hot in here, and the user is turning on the fan, and they might be able to drive better and might be less distracted because of that. Then how are you going to deal with them with empathy and give them the experience? Again, I would like to remind you that the camera sees these images. The AI sees these images. The AI is going to draw the conclusion of whether this is a distracted driver or not. So when you define the AI X design experience, it is important to have empathy for the user, desire on the values of the AI, and then create the experience and not believe blindly that the AI is saying the user is distracted or not distracted. Next, we will do a challenge using another image from the same dataset.

Contents