A Deep Learning Digitisation Framework to Mark up Corrosion Circuits in Piping and Instrumentation Diagrams

Abstract

Corrosion circuit mark up in engineering drawings is one of the most crucial tasks performed by engineers. This process is currently done manually, which can result in errors and misinterpretations depending on the person assigned for the task. In this paper, we present a semi-automated framework which allows users to upload an undigitised Piping and Instrumentation Diagram, i.e. without any metadata, so that two key shapes, namely pipe specifications and connection points, can be localised using deep learning. Afterwards, a heuristic process is applied to obtain the text, orient it and read it with minimal error rates. Finally, a user interface allows the engineer to mark up the corrosion sections based on these findings. Experimental validation shows promising accuracy rates on finding the two shapes of interest and enhance the functionality of optical character recognition when reading the text of interest.

A review of deep learning methods for digitisation of complex documents and engineering diagrams

“…The reviewed literature is listed by application and extracted data type in Table 1. Amongst these applications, there has been a considerable research focus on P&IDs (Rahul et al 2019;Sinha et al 2019;Yu et al 2019;Mani et al 2020;Gao et al 2020;Elyan et al 2020a;Moreno-García et al 2020;Jamieson et al 2020;Nurminen et al 2020;Paliwal et al 2021a;Moon et al 2021;Kim et al 2021b;Stinner et al 2021;Paliwal et al 2021b;Toral et al 2021;Bhanbhro et al 2022;Hantach et al 2021). Another research area is architecture diagram digitisation (Ziran and Marinai 2018;Zhao et al 2020;Rezvanifar et al 2020;Kim et al 2021a;Renton et al 2021;Jakubik et al 2022).…”

“…Most of the P&ID digitisation literature focussed on the extraction of specific data types (Sinha et al 2019;Gao et al 2020;Elyan et al 2020a;Jamieson et al 2020;Nurminen et al 2020;Moon et al 2021;Kim et al 2021b;Stinner et al 2021;Paliwal et al 2021b;Toral et al 2021). There is a particular focus on P&ID symbols (Elyan et al 2020a;Nurminen et al 2020;Paliwal et al 2021b).…”

“…One-stage object detection models have also been used for engineering symbol detection (Zhao et al 2020;Rezvanifar et al 2020;Elyan et al 2020a;Toral et al 2021;Zheng et al 2022). These models are faster than two-stage models.…”

“…Various deep learning models were used to detect text in complex diagrams, as shown in Table 3. The majority of studies used models designed for text detection, including Object detection models have also been used to detect text in engineering diagrams (Nguyen et al 2021;Hu et al 2021;Toral et al 2021). For example, Nguyen et al ( 2021) created a Faster R-CNN method to detect symbols and text in scanned technical diagrams.…”

Fields Touched by Digitalization: Analysis of Scientific Activity in Scopus

“…Publications on deep learning deal with harnessing neural networks to improve tasks such as computer vision, speech recognition and natural language processing. One manuscript on this topic is entitled "A Deep Learning Digitization Framework to Mark up Corrosion Circuits in Piping and Instrumentation Diagrams" prepared by [43], who present a semi-automatic framework that makes it easier for operators to load a piping and instrumentation diagram without going through digitization so that connection points and pipeline specifications are located through deep learning. Then, by means of a heuristic procedure, the text was obtained, oriented and read with the minimum degree of error, allowing the engineer to indicate the corrosion sections by means of a user interface.…”