Using Object Detection Technology to Identify Defects in Clothing for Blind People

Area of Science:

• Computer Vision
• Artificial Intelligence
• Assistive Technology

Background:

• Visually impaired individuals face significant challenges in identifying clothing defects such as stains or holes.
• Advancements in computer vision offer opportunities to mitigate these limitations and enhance clothing management for the blind community.

Purpose of the Study:

• To develop and evaluate an object detection system for categorizing and detecting stains on garments.
• To leverage the You Only Look Once (YOLO) architecture for automated defect inspection in clothing.

Main Methods:

• Collected a dataset of clothing defects, including various stains and conditions.
• Optimized the defect detection system through dataset expansion, data augmentation, and defect classification.
• Compared and evaluated three distinct YOLOv5 models for performance.

Main Results:

• The proposed YOLO-based system demonstrated high average precision (AP) in detecting garment defects.
• The approach proved effective across diverse and challenging defect detection scenarios.
• The system's performance indicates suitability for real-world applications.

Conclusions:

• The developed defect detection system is effective for assisting visually impaired individuals with clothing management.
• The research paves the way for a mobile application to enhance accessibility for the blind community.
• Object detection technology shows significant potential in creating assistive tools for daily living tasks.