Character Move analysis using OpenCV

Inspiration

I noticed a lot of sports broadcasts have heaps of data on a team or player and I though i could apply that to e-sports using computer vision l.

What it does

My application uses a sample library of moves and uses opencv to find the contours of these images. After some adjustment these data point ls are held for comparison againtst realtime moves. This allows the program to guess the move.

How I built it

I built it using a combination of opencv and pil. Opencv helped me find the contours which is necessary for comparison. Pil was used for video capture. With these tools and some clever manipulation of the mask I can create a method for comparing a moving characters states against the sample.

Challenges I ran into

I ran into standard desing issues and several geometric considerations. I limited the original scope because scaling the vectors proved harder than necessary for a first step.

Accomplishments that I'm proud of

I am proud that I got the program to guess correctly even if it was a trivial case.

What I learned

I learned to use opencv and pil. Likewise i learned a decent amount about the uses of opencv.

What's next for Character Move analysis using OpenCV

I hope to implement a machine learning to allow the machines guesses to be more refined. I also hope to find strageies to better handle backgrounds and scaling. Long term i hope to see it used to collect player statistics and maybe even used to create a computer player that operates like a player.

Built With

• opencv
• pil
• python