Analyzing Health Conditions of Critical Lift/Escalator Components with the Use of AI-Assisted Image Analytics

I&T Wish Analyzing Health Conditions of Critical Lift/Escalator Components with the Use of AI-Assisted Image Analytics
(REF: W-0452)
Matched I&T Solution
Trial Project
Summary and Challenges In this project, we aim to develop an AI engine to perform the full scrutiny of the site photos of some critical components of lifts/escalators for enhancing the efficiency and effectiveness of regulatory control, to replace the labor-intensive sample checking. Site photos collected from inspections, periodic examinations and special maintenances will be passed to the AI engine to perform classification and defect detection. Initially, 12 specific components should be classified and 2 specific components will be chosen for defect detection.

Project Challenges:
  1. Different angle, lighting condition, camera quality and photography skills shall be considered;
  2. No other data source will be provided or implemented except the captured site photos;
  3. Limited photos/videos could be provided for model training and thus certain techniques, e.g. data augmentation, transfer learning, synthetic data, active learning, and semi-supervised learning may be needed.
Expected Outcome
  1. Trained AI model(s) that can accurately classify the site photos into 12 specific components or exceptions;
  2. Trained AI model(s) that can provide defect detection for 2 components - Suspension Rope and Brake;
  3. An overall of 90% or above accuracy is preferred;
  4. The AI model shall be scalable and adaptable that can be applied new data for training and analysing via Application Programming Interface (API).
Expected Trial Duration 18-month
Contact Information
I&T Wish Proposer:Electrical and Mechanical Services Department (EMSD)
Contact Person:LAM CHAK KUI, Chokey
Position:Project Officer/Innovation/53
Tel:37418852
Email: lamchakkui@emsd.gov.hk
Upload Date 2023-04-21
Closing Date 2023-05-05