Vacuum Dirt Detection Brain

original test image of a clean floor
first image processing of clean floor
final image processing of clean floor
original test image of a dirty floor
first image processing of dirty floor
final image processing of dirty floor

The Inspiration came from one of the chapters of my current coursework. When I was first introduced to AI, I was taught about the basics of an AI vacuum cleaner and how it moves around. the part that was lacking during the coursework was how would the vacuum cleaner would know which location is dirty, hence my inspiration learn the basics of image processing, given an image of the floor, how would the program recognizes dirt vs clean

The program would take images of the floor (top view) as input, filter, and process the image, the goal is to recognize what part of the image is the dirt, and what part is the original floor. Hence, filtering out the floor in the image and whatever that is left will be the dirt

the program is built using python, using basic filters to separate the floor from dirt and process the image to recognize the horizontal and vertical tile grout line.

the main challenge was to first learn about image processing techniques. This is a difficult challenge for a student who does not know the topic. The information and high-level technique that takes years of research to attain are impossible to absorb in less than 36 hours. Aside from that, learning how to use functions that were never explored takes a bit of time to learn, and this small time spent eventually stacked up, eating up a major amount of time.

I was proud that the program was at least able to detect dirt on floor, still, I am disappointed in my methodology.

I have learned that image processing is very difficult. More work needs to be done

For the next improvement, definitely to be able to move to a more 3D image, then eventually able to distinguish furniture/shoes from dirt.