This week, I did the initial background subtraction from our dataset collected last week and also began trying to recognize some ingredients. One of the resulting videos is shown below. After a few frames, most of the static background is blacked out by the adaptive foreground mask, and we can see some red-outlined tofu squares as they are dropped in. Having the actual hot pot videos instead of random internet images was extremely helpful; based on our footage I realized it would be better to use OpenCV’s blob detector library rather than my original contouring approach.
I also worked extensively on creating and rehearsing the Design Review presentation for the upcoming week, which caused me to shift back my development schedule a little bit to accommodate the work going into the presentation. However, this time shift has not impacted any other components of the project so far, so we are predominantly on schedule. I hope to identify our remaining ingredients using blob detection in the coming week.