Encoder-decoder based active learning approach for corrosion segmentation in industrial and lab environments

Puntos clave

• Corrosion segmentation accuracy improves with active learning techniques in various environments.
• Performance threshold reached at 92% accuracy on segmentation tasks for both settings.
• Assessment using an encoder-decoder based model enhances detection efficiency in real-time applications.
• This method may enable better maintenance strategies, optimizing resource allocation in industry and labs.