# vball

**Repository Path**: liuheng0022/vball

## Basic Information

- **Project Name**: vball
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2020-09-20
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Ball tracking in volleyball

This repo is for ball recognition and tracking in live volleyball game.

## Requirements
- Python3
- OpenCV
- Keras with Tensorflow

## How to use

1. Get a video file with a game fragment
2. Get highest blobs:

*python3 high.py <path_to_vb_file> <mask_output_dir> <color_output_dir>*

3. Classify manually the blobs into 2 classes (b)all/(n)ot ball
4. Put the classified data into vball-net/train
5. cd vball-net
6. Python3 train.py
7. Test blobber: 

*python3 blobber.py <path_to_vb_file> 

You will see and output with ball paths like that:

<img src="https://github.com/tprlab/vball/raw/master/images/paths.jpg"/>

8. Run a player with the live ball tracking:

*python3 ball_play.py <path_to_vb_file> 

<img src="https://github.com/tprlab/vball/raw/master/images/vball_tracking.gif"/>

## Links

Used an open dataset from [some austrian league](https://www.tugraz.at/institute/icg/research/team-bischof/lrs/downloads/vb14/)