Deep Learning-Based Method for Detecting Traffic Flow Parameters Under Snowfall
2024

Detecting Traffic Flow Parameters in Snow

Sample size: 2300 publication 10 minutes Evidence: high

Author Information

Author(s): Jian Cheng, Xie Tiancheng, Hu Xiaojian

Primary Institution: Southeast University, Nanjing, China

Hypothesis

Can a deep learning framework effectively extract traffic flow parameters from videos during snowfall?

Conclusion

The proposed framework can accurately detect traffic flow parameters from snowy images, achieving a 97.2% accuracy under moderate snow conditions.

Supporting Evidence

  • The framework improves vehicle recognition accuracy by 8.6% after snow removal.
  • Traffic flow parameter estimation accuracy reaches 97.2% under moderate snow conditions.
  • The proposed method outperforms traditional traffic monitoring systems.

Takeaway

This study shows how to use smart computer programs to see cars and measure traffic even when it's snowing.

Methodology

The framework includes four stages: snow particle removal, snow streak removal, vehicle detection using yolov5, and traffic flow parameter estimation using the virtual coil method.

Limitations

The model's accuracy decreases with heavier snowfall due to increased snow density.

Digital Object Identifier (DOI)

10.3390/jimaging10120301

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication