Implementation of YOLOv5 and Vision OCR Hybrid Model for GD&T Recognition

The interpretation of Geometric Dimensioning and Tolerancing (GD&T) in engineering drawings is a critical aspect of design and manufacturing processes. However, traditional methods of manual annotation are time-consuming and often lead to variability in understanding, which can impact product functionality and inspection outcomes. To address these challenges, this paper proposes an automated approach to recognize GD&T in engineering drawings by combining deep learning techniques. The primary objective of this paper is to develop a hybrid model that integrates the YOLOv5 object detection model and Vision OCR for symbol, text, and character extraction. The purpose is to streamline the interpretation process and improve the accuracy of GD&T recognition in engineering drawings. By training the YOLOv5 model on a diverse dataset and employing Vision OCR for text retrieval, the model aims to detect objects and extract relevant text efficiently. Performance evaluation metrics, including precision, recall, and mean Average Precision (mAP), are used to assess the effectiveness of the proposed hybrid model. Experimental results demonstrate promising outcomes, with the model achieving high precision and recall rates, as well as a strong mAP score. These results indicate that the hybrid model can accurately recognize objects and text within engineering drawings up to 80%, thereby addressing the problem of inefficiency and variability associated with manual GD&T interpretation. This paper offers a novel solution to automate GD&T recognition in engineering drawings, contributing to enhanced efficiency and accuracy in design interpretation. The proposed model has significant implications for engineering graphics and design practices, as it facilitates better communication and collaboration among engineers, designers, and manufacturers. By streamlining design documentation processes, the hybrid model can be integrated into manufacturing workflows to improve productivity and quality assurance in engineering practices.