Feature extraction using Hough transform for solid waste bin level detection and classification

Area of Science:

• Computer Vision
• Artificial Intelligence
• Environmental Monitoring

Background:

• Effective solid waste management is crucial for urban sustainability.
• Manual monitoring of waste bins is labor-intensive and inefficient.
• Developing automated systems can optimize waste collection routes and reduce costs.

Purpose of the Study:

• To develop and evaluate an automated system for solid waste bin level detection and classification.
• To utilize image processing and machine learning for accurate waste monitoring.
• To assess the system's performance using established metrics.

Main Methods:

• Hough transform techniques for feature extraction and line detection.
• Feedforward Neural Network (FFNN) model for waste level classification.
• Receiver Operating Characteristic (ROC) graph and Area Under Curve (AUC) for performance evaluation.

Main Results:

• The FFNN model achieved high accuracy in classifying waste levels.
• The system demonstrated excellent performance for WS-class (AUC: 0.9875) and good performance for WS-grade (AUC: 0.8293).
• The developed system proves effective for solid waste bin monitoring.

Conclusions:

• The proposed image detection and classification system offers a viable solution for automated solid waste bin monitoring.
• The system's high accuracy and efficiency can benefit municipal authorities.
• This technology supports smarter waste management strategies and environmental sustainability.