For this week's assignment, I tried training a ML model to get some practice. I decided to teach it the song ‘head shoulders knees and toes’ to see if it'd be able to tell the difference between those body parts. I find it funny and also interesting to treat the ML model sort of like a little kid. I took between 120-160 pictures for each category in various different angles

After exporting it and putting it on the p5.js sketch, I was able to test it and see if it worked. The sketch seemed to run smoothly and it was very responsive to the body parts:

One funny thing that I noticed is that the model was also picking up the camera angles, so the response would change based on where the camera was pointing

When testing the model, I also tried disguising myself to see if the model would still pick up the ‘head’ portion and it did, which was a nice surprise

One other funny thing that happened was when Fatima tried to play the game, but the model would recognize her mostly as ‘knees’. I assume it's because of the color of her jacket, which is close to the color of my pants.

link to sketch here.