After receiving feedback from the previous week’s proposal, our team took a step backwards to redefine our idea this week. Earlier this week, I communicated with our faculty and TA to clarify their feedback. After validating their feedback and deciding to change ideas, I took charge of fleshing out our backup idea (use-case/challenges/solution/testing) of a camera system that does auto-tracking and auto-generation of stream of a toy car race. I also helped clean up our list of possible fouls to detect for our main idea.
After presenting everything to the staff on Wednesday and part of Thursday, we decided on pursuing the detection of double-hit fouls and push fouls in pool. Then the biggest challenge was determining what the required camera resolution would be to capture a double hit. So I spent some time in the University Center game room performing double-hits and recording them as iphone video(~30fps) or slowmo(~180fps) to see what the images would look like frame by frame. Attached is the frame by frame images (~180fps) of a double-hit.
Since there were only 1-2 frames between the two hits, and that the iphone slowmo feature is not quite 200fps, we wanted to see what 260fps footage would look like, so I purchased a 260fps camera. It came today so I will be trying it out tomorrow.
Our schedule has been delayed since our proposal week, so I hope to get back on track by testing out the 260fps camera and: 1. if camera footage is good, team can move forward to rely more on camera footage 2. if camera footage is not quite good, team can look more into using other aspects of the hit to determine if it was a foul or not. I will also be ordering the pool table and cameras this coming week.
Next week, I hope to order the parts needed to start obtaining footage of double hits. I also hope to have a plan to use frame by frame images to detect double hits using collisions as well as distanced traveled by the balls.