https://github.com/cbaezp/roulette
Proof of concept using a custom-trained vision model for object tracking . First step in exploring AI risks and use cases for online casinos.
https://github.com/cbaezp/roulette
computer-vision machine-learning roulette
Last synced: 3 months ago
JSON representation
Proof of concept using a custom-trained vision model for object tracking . First step in exploring AI risks and use cases for online casinos.
- Host: GitHub
- URL: https://github.com/cbaezp/roulette
- Owner: cbaezp
- License: mit
- Created: 2024-01-17T00:22:58.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-01-21T17:09:40.000Z (over 2 years ago)
- Last Synced: 2025-10-12T23:29:57.674Z (8 months ago)
- Topics: computer-vision, machine-learning, roulette
- Language: Python
- Homepage:
- Size: 25 MB
- Stars: 8
- Watchers: 1
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Roulette Video Processor
## Overview
This project is a proof of concept designed to explore the capabilities of computer vision in the context of object tracking. Using OpenCV and YOLO (You Only Look Once). The implementation includes `RouletteWheelTracker` and `RouletteBallTracker` for tracking the roulette wheel and ball, respectively.
## Prerequisites
- Python 3.10
- **Important:** PyTorch must be installed before proceeding with the installation of other dependencies. Visit the [official PyTorch website](https://pytorch.org/get-started/locally/) for detailed installation instructions.
## Installation
Clone the repository to your local machine:
```bash
git clone https://github.com/cbaezp/roulette
cd roulette
```
After installing PyTorch, install the remaining Python dependencies:
```bash
pip install -r requirements.txt
```
## Usage
To run the video processing script, navigate to the repository's root directory and execute:
```bash
python roulette.py
```
### Parameters
- `VIDEO_PATH`: Path to the input video file (e.g., `videos/roulette_test.mp4`).
- `OUTPUT_PATH`: Path for the output video file (e.g., `output_video.mp4`).
- Both, `RouletteWheelTracker` and `RouletteBallTracker` have additional parameters that could be updated based on the video, angle, etc.
### Customization
You can customize the behavior of the trackers by modifying their initialization parameters in `roulette.py`.
## Components
### VideoProcessor
Manages the video processing pipeline, handling frame reading and writing operations.
### RouletteWheelTracker
Detects and tracks the roulette wheel in video frames.
### RouletteBallTracker
Utilizes YOLO, powered by PyTorch, for the precise tracking of the ball's position and trajectory.
## Proof of Concept
This project demonstrates the practical application of computer vision techniques, showcasing object tracking under dynamic conditions.
## YouTube
[![Check out the youtube video]](https://youtu.be/bpy933SQ6Q0)