Team B9: BreakTime

Harry

I worked with Christina to tune values and work on integration with our HSV filtering and Edge detection algorithms. I worked with the team to collect our testing results for the final presentation, and I am now working to increase stability and reduce variance for our balls and cue stick for the final demo. Christina and I are working on really fine-tuning the cue stick detection.

Christina

I helped collect the CV testing data for our final presentation, and I spent time understanding the full project and rehearsing my Wednesday presentation. Prior to the final demo, I am mostly working on improving the consistency of our CV’s readings for a smooth user experience. Harry and I are working on getting the cue stick angle and position to be as accurate as possible.

Sam

To help collect test data, I built a physical testing setup to measure our software backend metrics. This helps us compare our software prediction with the true pool table physics. Additionally, Tamal suggested that we incorporate cue stick speed, so I created a cue stick speed module. I tuned the module and included PyGame functionality so that there is useful projector output of the cue stick speed.

Team

We worked a lot on metrics and gathering testing results for our final presentation. We ran tests on our CV component, measuring the difference between the physical and CV-computed locations of the pool balls. For the software backend, we created a testing setup for different angle shots and recreated that setup in the software. From this, we compared the actual ball angle to the software-predicted ball angle. We also created our end-to-end test suite and compared the location differences. After all of these tests, we all worked to create our final presentation and supported Christina’s talk.

Now, we are working on getting our project ready for the final demo. We want our system to be very consistent, function correctly in the gym demo space, and have an intuitive and usable interface for our users. Wish us luck in these final days!

TEAM

This short-week (because of Carnival), we all came together to make an alternative image perception algorithm worked. We employed and tested an edge detection algorithm with circle contours to detect the different pool balls. Additionally, we restructured how the camera would be attached to the frame so that the camera position is independent of the projector. We are experimenting with different combinations of pool ball detection to find the method with the most accuracy. Only 3 weeks left — let’s make it count! 😇

Christina

This past week, I worked on the design presentation and report outline with the rest of the team. Our 40″ pool table arrived on Monday, so we finally had access to the table that we will use on demo day. We were able to take photos and accurately measure a random layout of objects of our demo table. I modified the computer vision object detection for this new pool table and ball setup. The object detection is measuring all of the balls needed for 9-ball pool, and I will work on improving the accuracy of the ball center detections. One major complication was the green pool ball against the green felt of the table, and I hope to continue testing this distinction especially as our lighting conditions change. In the next week, I hope to compare the physical measurements we took with a ruler to the CV’s perceived object locations. I will try to fine-tune the detection parameters to minimize the difference between the actual and perceived pool ball locations.

Harry

Last week, I was the presenter for our design presentation, so I spent a lot of time preparing, finalizing, and rehearsing the presentation. I started working on the design of our frame and how we will use the PVC parts to build our sturdy frame. I researched most of the hardware parts we will need for our camera and frame setup, and we have placed the orders for them. When the camera came in, I helped Christina integrate the CV program with the video camera input. Along with the others, I worked on the design slides and documentation for the report. In the next week, the parts will arrive at Home Depot, so I can start prototyping and building the frame setup.

Sam

This past Sunday, I accompanied the team at one of our most productive meetings. It was a very windy day but we found save haven in one of the giant auditoriums at Tepper. Harry proceeded to kill the presentation the next day — sighs, podiums, and all. Later in the week, our 40″ pool table arrived and I played around with it. After shooting a few balls and trying different types of shots, I realized that if we end up using the 40″ pool table for our final demo, the physics is much simpler. For starters, there is no ‘english’ (spin) really possible on the ball, which makes things simpler. Additionally, the ball has to be hit at a minimum speed, or the balls will ‘break’ like golf balls on a green, due to the manufacturing errors in the table (not a flat surface). All-in-all, learned a lot this week and hopefully can move forward. Next week, I will wrap up my code for the MVP, refactor, and possibly work on integration with CV.

TEAM

As a team, we polished up our design presentation. We met together the day beforehand to finalize all the details and clarify any remaining topics. Once we received the design presentation feedback, we had a candid discussion on what went well and the different ways we could improve our system design. We followed the presentation directions well and were very prepared for any potential questions on our design. However, there were places we should have added in more detail and explained our reasoning for our decisions.

After receiving presentation feedback, we met on Thursday to start drafting our design report. We began with a brainstorming session to enumerate the components and workflow of our system.

We then started outlining our report, adhering to the final project report template. We have already collaborated and discussed the major points for our paper, and we are on track to finish our report Monday.