EA NHL Video analyzer

Prerequisites

Docker and Docker compose need to be installed in order to run the application.

Usage

Video analysis

The Tensoflow model required in order to use the classifier for video stream. This can be created by training or used a pretrained model. A pretrained model can downloaded from: https://drive.google.com/open?id=1jYaGyjpTR7gRpY4QQLzLQxbzh8RV5b2o

To run the video analyzer for a YouTube link, run:

docker-compose run --rm app ./video_analyzer.py https://www.youtube.com/watch?v=MjhNIDqOO5Q

To run the video analyzer for a local video file, run:

docker-compose run --rm app ./video_analyzer.py <video_path>

Output is an array of (event time, class)-tuples. Example:

[
  (1, 'game'),
  (2, 'game'),
  # ...
  (1000, 'goal'),
  # ...
  (1600, 'results'),
]

TODO: Do something useful with the classes.

Image classification model training

Training data expects a root path containing images in three adjacent folders named train val test. Example dataset can be downloaded from: https://drive.google.com/open?id=1hlDgZ2pVPAmMbXuxpyrgv-i9sUF89Q7U

Run training:

docker-compose run --rm app ./image_classifier.py train <data_path>

Data extractor for results page

Use docker-compose to run the extractor:

docker-compose run --rm app ./stat_extractor.py examples/NASC_Turku_21-4_screenshot.png

This outputs the extracted stats into a csv file to out directory. Example output:

Stat,Away,Home
Team,DAL,COL
Goals,2,4
Shots,20,19
Hits,99,15
Time on attack,06:09,08:11
Passing,58.6%,76.4%
Faceoffs won,10,15
Penalty minutes,08:00,04:00
Powerplays,1/2,1/4
Powerplay minutes,01:48,05:29
Shorthanded goals,0,00

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Please make sure to update tests as appropriate.

License

MIT