Machine Learning Video Analysis to Assess Pitcher’s Injury Risk in Baseball Games
Abstract:
Bibliography/Citations:
Danny, Malter Analytics, https://malteranalytics.github.io/mlb-openpose/
Z. Cao, G. Hidalgo Martinez, T. Simon, S. Wei, and Y. A. Sheikh. Openpose: Realtime multi-person 2d pose estimation using part affinity fields. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019.
Zhe Cao, Tomas Simon, Shih-En Wei, and Yaser Sheikh. Realtime multi- person 2d pose estimation using part affinity fields. In CVPR, 2017.
Tomas Simon, Hanbyul Joo, Iain Matthews, and Yaser Sheikh. Hand key- point detection in single images using multiview bootstrapping. In CVPR, 2017.
Shih-En Wei, Varun Ramakrishna, Takeo Kanade, and Yaser Sheikh. Convolutional pose machines. In CVPR, 2016.
Additional Project Information
Project files
Research Plan:
First, I'll download Major League Baseball (MLB)'s pitching videos via baseballsavant.mlb.com and organize them by pitchers. The videos are then cropped and cut for processing efficiency.
Second, I'll use the open-source OpenPose algorithm will be used to detect the pose of pitchers' pitching motion in each video and create movement graphs.
Third, I'll gather pitcher injuries through MLB's public record of injury list.
Last, I'll train a neural network to find correlation between different pitching motions and injury list stays.
Questions and Answers
1. What was the major objective of your project and what was your plan to achieve it?
My main objective is to find correlations between certain pitching patterns and injuries, thus creating a method to predict and prevent future injuries.
a. Was that goal the result of any specific situation, experience, or problem you encountered?
I'm an avid baseball fan and had witnessed the depression of many professional pitcher injuries. I hope that this study can help mitigate the rampant pitching injuries.
b. Were you trying to solve a problem, answer a question, or test a hypothesis?
I'm trying to solve the problem of pitching injuries.
2. What were the major tasks you had to perform in order to complete your project?
My major tasks include downloading and processing videos, implementing OpenPose algorithm, and training neural network.
3. What is new or novel about your project?
a. Is there some aspect of your project's objective, or how you achieved it that you haven't done before?
Prior to this project, I've never used the OpenPose algorithm. I also never training neural network of such large scale.
b. Is your project's objective, or the way you implemented it, different from anything you have seen?
I've seen machine learning video analysis implemented on baseball videos, but my research is distinct in that it uses the analysis to analyze injury risks. I've also seen other studies about baseball injury risks, but they either require in-person data collection or doesn't involve machine learning.
c. If you believe your work to be unique in some way, what research have you done to confirm that it is?
I've read relevant research papers, and my methodologies are unique.
4. What was the most challenging part of completing your project?
a. What problems did you encounter, and how did you overcome them?
I encountered the problem of inconsistent video quality and imprecision in OpenPose results. I overcame these difficulties by manually pruning the dataset. I also encountered the problem of camera angles being different between videos. To solve this problem, I grouped the pitching videos by calculating location and angles of pitcher stance to analyze pitching pattern with greater precision.
b. What did you learn from overcoming these problems?
I learned varied techniques in computer science research. I also learned the importance of consulting with classmates and teachers, who often brings great insights to my problems.
5. If you were going to do this project again, are there any things you would you do differently the next time?
I would try various machine learning motion capturing models and compare their results. If possible, I would also prefer data from in-person experiment.
6. Did working on this project give you any ideas for other projects?
The same methodology can be applied to other sports with high injury risks, such as football.
7. How did COVID-19 affect the completion of your project?
None.