Title:
Enhancing Semantic Processing Of Single-line Diagrams Using Hybrid Deep-learning And Bayesian Reasoning Methods

Author:

Supervisor:

Department:

Computing

Year:

2025

URL:

Abstract:

Accurate semantic processing of single-line diagrams (SLDs) is essential for analyzing and understanding electrical systems, as they encapsulate critical information about operational safety and effic...
View more
Abstract Image

Multimedia Content:

01

Methodology

1. Preliminary Pilot Experiment Conducted on a small dataset (600 SLD images) to test feasibility. Steps included: • Data Collection & Preprocessing: Grayscale conversion and text removal to reduc...
View more
Methodology Image
02

Citation Analysis

1. Advanced Engineering Informatics (Elsevier) Scope: Engineering informatics; knowledge‑intensive, AI/ML methods applied to engineering tasks (BIM, design knowledge extraction, computer vision for...
View more
03

Impact

1. Scientific Impact Advances Semantic Processing of Electrical Diagrams: Introduces a novel deep learning model with hybrid residual attention and proximity-aware loss, improving symbol classifica...
View more
04

Benefits to Society

1. Technical Benefits Higher Accuracy in Diagram Interpretation: Achieves 92.5% classification accuracy and reduces localization errors by 17%, improving reliability in symbol recognition. Faster ...
View more
05

Applications

1. Automated SLD Symbol Classification Application: Real‑time identification and labeling of electrical symbols in SLDs. Use Case: Bulk processing of legacy PDF/TIFF drawings to auto‑tag transform...
View more
06

Infographic

Summary of this research
Infographic Image
07

Patents / Products

1. Intelligent Electrical Diagram Interpretation System Patent: A system for automated interpretation of Single-Line Diagrams using a hybrid deep learning architecture with residual attention and p...
View more
08

Experts

1. Dr. Yew Kwang Hooi Institution: Universiti Teknologi PETRONAS (UTP), Malaysia Contribution: Main supervisor; provided strategic guidance on deep learning architecture design and Bayesian reason...
View more