Road Signs & Machine Learning: Day 1

By Abhi Mora

Setting Up the Project

Today, I dug a little into how Artificial Intelligence can detect the signs in the first place. This is where I learned about the type of AI used in Self-Driving cars, machine learning. What machine learning does is that rather than the user making defined rules, the computer will find similarities in different objects no matter the lighting, angle, etc. This astonished me, as it seems dangerous that we have to rely on computers to make the rules, which means that these self-driving cars have had minimal human involvement, with humans only supervising the program.

After I realized this, I wanted to know if I could see the code that the Machine Learning makes based on data. To do this, I went outside and got ten pictures of "stop" signs and ten pictures of "no outlet" signs. Rather than getting the same shot, however, I used different stop signs and angles, taking some while the sun was still there, and others while the sum was diminishing. Here are some images examples:

Angle Side: Left

Lighting: Darker

Stop Sign: Number 1

Angle Side: Center

Lighting: Lighter

Stop Sign: Number 1

As I conclude the day, not only do I have a lot of data to use in the next few days, but I also gained a lot of background knowledge. I am much more intrigued about how Machine Learning can think by itself, and if this will be the next global issue that needs to be addressed. The amount of technology we have already has surprised me, and I believe that the growth in technology will continue for many years and counting.