LyFormer based object detection in reel package X-ray images of semiconductor component

Abstract

With the development of artificial intelligence (AI) technology, companies are rationalizing the facilities required at the production site to suit smart factory plants, and applying AI to the production and inspection processes. In terms of manufacturing production, AI-computer vision can replace existing rule-based systems and can add competitiveness to industrial sites. In order to respond to the innovative business management paradigm in the manufacturing industry, the advancement of smart factory construction in the domestic manufacturing industry is in progress. Accordingly, the introduction of manufacturing execution systems (MES) that apply AI to industrial sites is becoming important. In the context of manufacturing lines that operate under surface mount technology (SMT), the precise quantification of electronic components is crucial for on-time supply chain management and effective production output. However, small-to-medium enterprises often grapple with the lack of sophisticated systems for accurate part counting. Traditional X-ray machines operate on rule-based algorithms, engendering discrepancies in component numbers and engendering inefficiencies. Small object detection using AI deep learning is a useful technology in the production field and can be applied to various fields in the future. YOLOv5 is a fast, high-performance, one-stage object detection program. It uses pytorch, which is lighter than existing models and can be accessed easily by users. To better recognize small objects, we proposed the LyFormer model (LCTC, YOLO, Transformer), which added 1 layer to the existing YOLOv5 network, added 1Head, inserted a transformer module, and preprocessed the data with label normalization, correlation, local texture, and context feature maps. The proposed model is an improved model with better accuracy and speed than the existing YOLOv5 model. We took X-ray images of semiconductor parts, trained using the improved proposed model, and obtained excellent performance. It can be applied to the inspection process in the surface mount technology industry with many small devices in conjunction with production for the advancement of MES. The mAP performance of the improved proposal model was 0.672 compared to 0.399 of the YOLOv5 model, which was a significant improvement. The accuracy of the improved model was 0.915, which was a significant improvement compared to the result of the YOLOv5 model (0.602), and it could be applied directly in the field for object inference, as it could detect objects in reel units.

Keywords

• Small object detection
• Semiconductor
• X-ray
• YOLOv5
• Artificial intelligence
• LCTC