Deep Learning Detection of Defect-Free Bangles: Machine Vision Approach for Quality Assurance in Bangle Manufacturing

This study explores the application of deep learning and machine vision methods to identify defect-free, circular bangles. The main objective is to improve quality control in the manufacturing of bangles. The study introduces a new model that achieves a remarkable mean Average Precision (mAP) score of 99.1%, showcasing its exceptional accuracy and efficiency in the identification and categorization of bangles. The model successfully differentiates between “good,” “defect,” and “broken” bangles, as evidenced by thorough testing and analysis. During the testing phase, the model regularly achieved a 99% average detection rate for recognizing “good” bangles, hence confirming its capacity to properly detect perfect bangles. Moreover, the model demonstrated remarkable competence in detecting “defect” bangles, achieving detection rates ranging from 98% to 97%, thereby emphasizing its efficacy in recognizing flawed goods. During the analysis phase, the model consistently and precisely recognized “defect” and “broken” bangles, achieving an average detection rate of 98%. This confirms its capability to effectively classify various types of bangles. The model's high mean average precision (mAP) value, along with its exceptional testing and analysis results, highlights its promise as a reliable and effective solution for quality control in bangle manufacturing applications. This study highlights the substantial advantages of utilizing machine vision and deep learning for accurate quality assurance in bangle manufacturing.