CSCI-4962 Final Project

Hey!

Instructions to run the code:

Requires numpy, opecv-python, imutils. These can be installed by:

pip install numpy
pip install opencv-python
pip install imutils

To run:

python yolo_opencv.py -i hockey-1.png -c yolo.cfg -w yolov3.weights -cl classes.txt

python yolo_video_opencv.py -i cv-hockey-1.mp4 -y .\ -o test.mp4

Training Models

https://medium.com/@manivannan_data/how-to-train-yolov3-to-detect-custom-objects-ccbcafeb13d2

https://medium.com/coinmonks/detecting-custom-objects-in-images-video-using-yolo-with-darkflow-1ff119fa002f

./darknet detector train cfg/player6-obj.data cfg/player6-yolov3-tiny.cfg darknet53.conv.74

Related Research/Work

Github: BSI Vision

• "an open source platform that can injest broadcast footage of hockey games and output player mappings as well as other useful information harvested from these frames."

Fast and Reliable Detection of Hockey Players

Self-Learning for Player Localization in Sports Video

Learning to Track and Identify Players from Broadcast Sports Videos

Automatic Acquisition of Motion Trajectories: Tracking Hockey Players

Pose Estimation of Players in Hockey Videos using Convolutional Neural Networks

Classification of Puck Possession Events in Ice Hockey