This week, I fleshed out the vision of what our final product for Hot Pot Bot might look like (shown in the sketch below), and defined several timing and user interaction requirements for the system. I also did the initial setting up for our team’s online Gantt chart. As I will be mainly in charge of writing the computer vision aspect of our project, I spent a lot of time this week making sure it would be feasible. I narrowed down our ingredient categories to five common hot pot ingredients with shapes and colors that, based on prior experience with computer vision, I am confident an algorithm could distinguish well. Fish balls and shiitake mushroom caps can be categorized by their rounded shape and distinguished by their color. Tofu can be identified by its square shape, and meat and bok choy can be identified by their red and green colors, respectively.
Since our timeline has just begun, we are on schedule with all of our deliverables for now. Next weekend, our team will film videos of hot pot for use in developing our ingredient identification algorithm. During the time in between, I plan to do make some preliminary code sketches to see what kind of contours and color information I can get from images of our chosen ingredients on the internet.
