Application of Deep Learning in Civil Engineering Management

Area of Science:

• Civil Engineering
• Computer Vision
• Machine Learning

Background:

• Construction safety is critical in civil engineering.
• Worker safety, particularly helmet detection, is a key concern.
• Current methods for safety monitoring require improvement.

Purpose of the Study:

• To develop an intelligent system for recognizing construction safety clothing.
• To improve the performance of construction safety helmet recognition.
• To enhance the reliability of safety monitoring in construction sites.

Main Methods:

• Utilizing deep residual networks (ResNet) and sparse representation-based classification (SRC) as base classifiers.
• Implementing a fusion strategy for decisions from ResNet and SRC.
• Employing a method to update classifiers with reliable test samples for improved accuracy.

Main Results:

• The proposed method demonstrates effectiveness and robustness in recognizing construction safety clothing.
• Experimental results validate the performance using actual measured data.
• The system achieves reliable recognition of safety gear.

Conclusions:

• The combined deep learning and traditional classifier approach significantly improves safety clothing recognition.
• The proposed method offers a robust solution for intelligent safety monitoring in construction.
• Continuous learning through sample updates enhances recognition accuracy over time.