Research on Implementation of BADGE Active Learning to Increase Defect Detection Effectiveness

Defect detection is a critical task in industries ranging from manufacturing process to quality control. It is a great investment to automate the process to increase productivity. Automating the defect detection process using deep learning such as the YOLO (You Only Look Once) algorithm has shown remarkable performance in object detection tasks. Further integrating the YOLO algorithm with BADGE (Batch Active learning by Diverse Gradient Embeddings) active learning algorithm has the potential to enhance the learning process of the model. This integration of BADGE with YOLO aims to increase the productivity by reducing the labeling effort. This study explores the methodological insights and the challenges encountered during the research process of integrating BADGE active learning with the YOLO algorithm. Furthermore, the study discusses potential solutions and future studies to address the encountered challenges.