JoVE - Journal of Visualized Experiments
JoVE Visualize
Contact Us

RFGLNet for adverse weather domain-generalized semantic segmentation with frequency low-rank enhancement

  • 0Xi'an Technological University, Xi'an, China.
Scientific Reports +

|

February 10, 2026

|

February 10, 2026

View abstract on PubMed

Summary

This summary is machine-generated.

This study introduces RFGLNet, a new model for semantic segmentation in adverse weather. It achieves high accuracy in challenging conditions like rain and fog, crucial for autonomous driving.

Area Of Science

  • Computer Vision
  • Artificial Intelligence
  • Robotics

Background

  • Semantic segmentation in adverse weather is challenging due to poor visibility and noise.
  • Existing methods struggle with object details and global context in severe conditions.
  • Domain generalization (DG) aims to improve model robustness in unseen environments.

Purpose Of The Study

  • To develop a domain-generalized semantic segmentation model for robust performance in adverse weather.
  • To enhance the capture of object details and global structures in challenging conditions.
  • To enable reliable perception for autonomous driving systems.

Main Methods

  • Introduced RFGLNet, incorporating SVD-Initialized Low-Rank Module, Fourier-Enhanced Channel Attention, and Grouped Modeling Spatial Attention.
  • Leveraged frequency-domain information via Fourier transforms for improved global perception.
  • Employed singular value decomposition (SVD) for efficient parameter fine-tuning.

Main Results

  • RFGLNet achieved a mean intersection over union (mIoU) of 78.3% on the ACDC adverse weather dataset.
  • The model demonstrated improved segmentation accuracy in challenging conditions.
  • RFGLNet required only 4.32 million trainable parameters, indicating parameter efficiency.

Conclusions

  • RFGLNet offers a robust solution for domain-generalized semantic segmentation in adverse weather.
  • The proposed modules effectively enhance global and local feature extraction.
  • The model shows significant promise for enhancing the safety and reliability of autonomous driving.

Keywords:

Related Concept Videos

What is Weather? 01:07

20.3K

Overview

Weather refers to the current state of atmospheric conditions at a specific time and place. Weather is temporary and is influenced by the interactions between the sun, atmosphere (gases around the Earth), hydrosphere (the water bodies of the Earth), and geosphere (the land portion of the Earth). As all of these factors differ across location and time, different weather can be observed across the planet.

Solar Radiation and the Greenhouse Effect

Solar radiation penetrates the...

Linear Approximation in Frequency Domain 01:26

386

Linear systems are characterized by two main properties: superposition and homogeneity. Superposition allows the response to multiple inputs to be the sum of the responses to each individual input. Homogeneity ensures that scaling an input by a scalar results in the response being scaled by the same scalar.
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....

Frequency-Domain Interpretation of PD Control 01:24

394

Proportional-Derivative (PD) controllers are widely used in fan control systems to improve stability and performance. A fan control system can be effectively represented using a Bode plot to illustrate the impact of a PD controller through its transfer function. The Bode plot visually conveys how PD control modifies the fan's response across various frequencies, providing a frequency domain interpretation of the controller's behavior.
The proportional control gain, combined with the...

Time and frequency -Domain Interpretation of PI Control 01:27

439

Proportional-Integral (PI) controllers are essential in many control systems to improve stability and performance. They are commonly used in everyday devices like thermostats to enhance system damping and reduce steady-state error. When the zero in the controller's transfer function is optimally placed, the system benefits significantly in terms of stability and accuracy.
Acting as a low-pass filter, the PI controller slows the system's response and extends settling times. This requires...

Time and frequency -Domain Interpretation of Phase-lead Control 01:24

479

Phase-lead controllers are commonly used in various control systems to enhance response speed and stability. Adjusting the brightness on a television screen offers a practical example of phase-lead control. When contrast is enhanced, a phase-lead controller is employed. Mathematically, phase-lead control is identified when the first parameter is smaller than the second.
The design of phase-lead control involves the strategic placement of poles and zeros to balance steady-state error and system...

Time and frequency -Domain Interpretation of Phase-lag Control 01:21

423

Phase-lag controllers are widely used in control systems to improve stability and reduce steady-state errors. A dimmer switch controlling the brightness of a light bulb serves as a practical example of phase-lag control, gradually adjusting the bulb's brightness. Mathematically, phase-lag control or low-pass filtering is represented when the factor 'a' is less than 1.
Phase-lag controllers do not place a pole at zero, but instead influence the steady-state error by amplifying any...