Research on OCR Detection Technology for Beer Bottle Cap Inkjet Coding in Industrial Scenes Based on Deep Learning

The inspection of inkjet coding on the filling line in the beer industry plays a crucial role in ensuring the quality control of products before shipment. Currently, the primary method in industry relies on manual inspection, which suffers from defects such as waste of manufacturing resources, low efficiency, and a high error rate. Against this backdrop, this article proposes an OCR (Optical Character Recognition) model based on deep learning, including image processing, improved DBNet for text detection, a direction classifier for text box correction, and an improved CRNN for text recognition. This method achieves over 90% accuracy, significantly improving inkjet coding inspection efficiency on beer bottle caps in industrial settings.