A Hybrid Deep Learning and Improved SVM Framework for Real-Time Railroad Construction Personnel Detection with...

Area of Science:

• Civil Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Railroad construction sites pose significant safety risks.
• Current manual and image processing methods for personnel monitoring lack real-time accuracy.
• Effective safety monitoring is crucial for accident prevention and construction efficiency.

Purpose of the Study:

• To develop an advanced railway worker detection method for improved safety and efficiency.
• To overcome the limitations of traditional monitoring techniques in complex construction environments.
• To enhance the accuracy and real-time performance of personnel detection systems.

Main Methods:

• Utilized non-local mean noise reduction and histogram equalization for image pre-processing.
• Extracted multiscale features using Inception v3 and applied Principal Component Analysis (PCA) for dimensionality reduction.
• Employed an improved Support Vector Machine (ISVM) classification algorithm for personnel detection, incorporating data enhancement and K-fold cross-validation for small sample optimization.

Main Results:

• The ISVM method demonstrated significant improvements in detection accuracy.
• The proposed approach achieved enhanced real-time performance compared to traditional methods.
• The system effectively addressed personnel detection challenges in complex railroad construction environments.

Conclusions:

• The ISVM-based method offers a robust solution for railroad construction safety monitoring.
• This approach provides crucial technical support for enhancing worker safety and operational efficiency.
• The study validates the effectiveness of advanced AI techniques in high-risk industrial settings.