Dartboard Detection System

Overview

This repository contains the implementation of a dartboard detection system utilizing a combination of Viola-Jones object detection and advanced shape detection techniques. The project focuses on accurately identifying dartboard locations in images using cascaded classifiers and geometric shape analysis.

Project Structure

• Dartboard/
  • Contains essential data and scripts for the dartboard detection.
• dartboard_detector.py
  • Main Python script for dartboard detection using Viola-Jones and shape detectors.
• Dartboardcascade/
  • Directory containing the trained model files for the cascade classifier.
• groundTruth.txt
  • Text file containing the ground truth data for the dartboards in the images.
• negatives/
  • Folder with images used as negative samples during the training of the classifier.
• negatives.dat
  • Data file listing the negative samples.
• report.pdf
  • Detailed report documenting the methodology, results, and analysis of the dartboard detection system.
• template.jpg
  • Template image used for feature matching in the detection process.

Detection Pipeline

1. Viola-Jones Cascade Classifier: Initial dartboard detection using Haar features to filter potential dartboard regions.
2. Shape Detection:
   • Line and circle detection with Hough Transform to confirm dartboard features.
   • Integration of detected shapes to refine dartboard localization.
3. Feature Matching:
   • Utilizes FLANN based matcher for robust feature matching, overcoming issues related to object orientation and scale.
   • Combination of feature matching points and geometric shapes to finalize dartboard detection.

Highlights of Implementation

• Reduction of False Positives: Significant decrease in false positives through multi-stage filtering.
• Integration with Shape Detectors: Enhances detection accuracy by combining Viola-Jones detections with line and circle indicators from Hough Transform.
• Improvement Strategies:
  • Shift from template matching to feature matching to accommodate various dartboard sizes and orientations.
  • Weighted scoring system for potential dartboard regions to optimize detection accuracy.

Running the Detector

To run the dartboard detection system, execute the dartboard_detector.py script:

python dartboard_detector.py