Digitalization of Engineering Drawings - BC-403

Preferred Disciplines: Computer Science, Machine learning, Image processing, Graphics recognition (Masters, PhD or Post-Doc)
Project length: 2 years (6 units)
Approx. start date: As soon as possible
Location: Vancouver, BC or Toronto, ON
No. of Positions: 2
Preferences: SFU / UBC; University of Toronto, or University of Alberta
Company: Intelligent Project Solutions

About Company:

Intelligent Project Solutions (IPS) is a startup company based in Vancouver providing AI solution to meet our customers’ Engineering and Digitalization needs with a focus on Chemical/Oil/Gas and other process industries. We provide “Engineering with Intelligence”. Company is founded by industry expertise and data scientist from acadamy. Our Mission is to apply A.I technology and innovation for delivering faster, cheaper and better quality solutions for Engineering and Digital Transformation. Our long term Vision is to revolutionize Engineering with Artificial Intelligence

Summary of Project:

Huge amount of Engineering information are left outdated and none digitalized to be useful; and, huge amount of none value added work still exist in the current Engineering and digitalization process in Oil, Gas and Chemical industry. The pressing benefits and needs for digitalization of oil, gas and chemical industry, and needs for “faster, cheaper and better” capital project delivery creates immediate market needs for better industry solutions. Intelligent Project Solution, IPS, at this stage is focusing on “digitalization of Engineering drawings”. The projects IPS proposed are aimed to develop AI based solution to digitalize various Engineering Drawings. The existing open source platform and academic research do not meet the needs and challenges the issues presented.

 Research and development needs for meeting specific challenges in this area as described in the research objectives.  

Research Objectives/Sub-Objectives:

Develop workable methdology and/or algorithm to address each challenges of  graphics recognition issues in complex engineering drawings (CED):

  • Symbols or components in CED are numerous and are too similar to be differentiated; Need to develop  methods and techniques for  symbol recognition and identification to differentiate these symbols and components while improving the accuracy; Text recognition will be a subset of this process.
  • After identification in previous step there will be a need to develop a logical framework describing relationship and connectivity of identified components on the drawing. The framework will extract information from CED and categorize them under symbols, shapes, measurements, notes, etc. . This logical framework will establish relationships among components which are critical for the digitalized information to be integrated in the current  digitalization software.
  • A generic framework needs to be developed to address different subsets of Complex Engineering drawings viz.P&ID, layout, equipement drawings. They all have different symbols/components, and possess unique logical relationship among them. The frame work will integrate digitalization for each subset by drawing upon the basic technology from above points.- This will lead the project into minimal two subprojects: Layout, equipment drawings digitalization.

Methodology:

    1. All the existing methodologies in research and academia related to this project draws upon techniques in image processing, graphics recognition, and computer vision. Recently some work has been done using custom heuristics and Support Vector Machine (SVM) classifiers.
    2. Given the ever-growing advancement in Deep learning and Artificial Intelligence we would like to use  a mix of intelligence based approach and image processing techniques. 

    Expertise and Skills Needed:

    • Sound Computer Science background and technical system software concepts.
    • Focused Python programming experience with its various domain libraries Pandas, NumPy, SciPy, Matplotlib, scikit-learn, SymPy, etc. is highly recommended.
    • Experience in machine learning, deep learning, especially CNNs is highly required.
    • Experience in OpenCV, OpenGL, Python libraries for deep learning Keras is highly required and with Theano, Tensorflow, Caffe would be appreciated.
    • Hardworking, dedicated, punctual and pssionate about Python coding with documentation.
    • Excellent communication skills, avid learner, team player, analytical thinking and highly adaptable.

    For more info or to apply to this applied research position, please

    1. Check your eligibility and find more information about open projects.
    2. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform
    Program: