Learning-Based Image Damage Area Detection for Old Photo Recovery.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Digital Image Processing

Background:

• Manual and semi-automatic methods for old photo repair are time-consuming and labor-intensive due to manual damage marking.
• Existing fully automatic methods lack control over damaged area detection, risking the preservation of historical photos.

Purpose of the Study:

• To propose a deep learning-based architecture for the automatic detection of damaged regions in old photographs.
• To develop a damage detection model that accurately identifies and marks damaged areas for subsequent repair.

Main Methods:

• A novel deep learning architecture was designed for automated damaged area detection in old photos.
• The model was trained and evaluated on its ability to identify complex damage patterns.

Main Results:

• The proposed model effectively detects complex damaged areas in old photos automatically.
• Damage marking time was reduced to less than 0.01 seconds per photo, substantially accelerating restoration workflows.

Conclusions:

• The developed deep learning model offers an efficient and effective solution for automatically detecting damaged areas in old photos.
• This automated approach significantly reduces the time and effort required for old photo restoration, improving preservation outcomes.