Electronic component detection with deep learning

The recognition of electronic components is of great importance in various fields such as industrial automation, quality control and waste management. Accurate and efficient identification of electronic components is required to optimise production processes, increase product quality and improve maintenance procedures. The aim of this research was to evaluate the effectiveness of new object detection models for electronic component recognition. A comprehensive comparative analysis was performed on several models, including different versions of the YOLO algorithm, such as YOLOv5 and YOLOv8. To evaluate the object recognition models, rigorous training and tuning was performed for each model on a large dataset containing electronic components of different sizes, types and orientations. The research results comprehensively demonstrate the advantages and disadvantages of each model in the field of electronic component recognition. This work improves the understanding of object recognition techniques by providing a valuable tool for electronics industry professionals looking for the most appropriate model for specific applications. As a result of the study, the YOLOv8x model achieved a precision of 0.948, mAP50 of 0.917, mAP50-95 of 0.828 and F1 score of 0.878. In addition, a sensitivity value of 0.843 was obtained using the YOLOv8m model. These results show that the proposed models are effective in the task of electronic component recognition.