Industrial_components_Classification

Numerous studies on object recognition and classification have contributed to improved object identification accuracy. Most of them have two features, that is, an extensive dataset and a complex architecture. However, regarding the shortage of dataset in some particular industrial fields, these current classical models cannot help solve some practical issues. This study focuses on warehouse receiving management. The reason is that the human entry error at the warehouse receiving stage still exists in the warehouse receiving management, and industrial components classification in this domain has less attention in the academic area. Therefore, a model used to classify the industrial components is needed to develop and fill this gap area. This paper generated a straightforward machine learning model to classify various industrial components. A collected mini image data set introduced a template image library to improve industrial components classification and count accuracy rate by comparing the variance between input test images and the pre-defined template image data sets. The white histogram correlation coefficient was chosen to be the best algorithm to classify these industrial components while conducting RGB colour comparison experiments. Besides, this proposed model classification experiment achieved a higher classification accuracy rate. Specifically, the authors proposed a classification and counting model application in the warehouse receiving management. The final model performance experiment results, classification accuracy rate, reached 91.37%. The counting accuracy rate achieved 94%, which demonstrated the model effectiveness and stability, expecting to be applied in the receiving management to reduce human entry error. Images and code are provided in Github.