Product Name:
Soccer Offside Detection System
Mission Definition:
Automatically detecting offside from game videos with visual based tools
Target Customer:
Soccer referees or football associations
Running command:
python3 track_game_offside_detection.py -i <input video> -o <result video>
Track_game_offside_detection.py is the python file that track the soccer game :
- Detecting players:
For detecting the players we used Tensorflow Object Detection API.We used the pre trained model over the COCO dataset. COCO data set is a large-scale object detection dataset.
-
COCO data set properties :
- Object segmentation
- Recognition in context
- 330K images (>200K labeled)
- 1.5 million object instances
- 80 object categories
- Object segmentation
- 91 stuff categories
-
In our project we care about 2 classes only :
- person
- sports ball
-
Examples :
Our code is based on the next website:
https://towardsdatascience.com/analyse-a-soccer-game-using-tensorflow-object-detection-and-opencv-e321c230e8f2
- Detecting soccer ball:
In order to detect the soccer ball that it is a harder task we use the next 2 methods in this order(this means that when the a method failed to detect the soccer ball we are trying to detect it using the next method):
1.SSD_mobilenet_COCO model(more details above)
- Examples :
2.Motion tracking
The motion tracking algorith is based on the next project with modifications to detect a soccer ball:
https://www.pyimagesearch.com/2015/09/14/ball-tracking-with-opencv/
- Examples :
track_game_offside_detection.py combines the result of these 2 methods for detecting the soccer ball.
We implemented a Template matching algorithm too but we remove it as we did not get better results with this algoritm. The template matching algorithm is based on the link below:
https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_template_matching/py_template_matching.html
