Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ProvRain: Rain-Adaptive Denoising and Vehicle Detection via MobileNet-UNet and Faster R-CNN

Aswinkumar Varathakumaran, Nirmala Paramanandham

TL;DR

The paper tackles nighttime vehicle detection under adverse weather by introducing ProvRain, a lightweight denoising-augmented pipeline designed for real-time automotive use.ProvRain integrates a weather-aware MobileNet-U-Net denoiser trained via curriculum learning with a light proposal generator and a Faster R-CNN classifier to robustly detect headlights/taillights and suppress rain-induced noise.Key contributions include the curriculum-based denoising framework, a light-weight perception stack suitable for automotive hardware, and a multi-stage fusion of geometric, temporal, and classification cues that enhances early warning and tracking under rain.Empirical results on the PVDN dataset show improved accuracy, recall, PSNR/SSIM metrics for denoising, and superior early-warning performance compared with baseline methods and heavier transformer-based denoisers.

Abstract

Provident vehicle detection has a lot of scope in the detection of vehicle during night time. The extraction of features other than the headlamps of vehicles allows us to detect oncoming vehicles before they appear directly on the camera. However, it faces multiple issues especially in the field of night vision, where a lot of noise caused due to weather conditions such as rain or snow as well as camera conditions. This paper focuses on creating a pipeline aimed at dealing with such noise while at the same time maintaining the accuracy of provident vehicular detection. The pipeline in this paper, ProvRain, uses a lightweight MobileNet-U-Net architecture tuned to generalize to robust weather conditions by using the concept of curricula training. A mix of synthetic as well as available data from the PVDN dataset is used for this. This pipeline is compared to the base Faster RCNN architecture trained on the PVDN dataset to see how much the addition of a denoising architecture helps increase the detection model's performance in rainy conditions. The system boasts an 8.94\% increase in accuracy and a 10.25\% increase in recall in the detection of vehicles in rainy night time frames. Similarly, the custom MobileNet-U-Net architecture that was trained also shows a 10-15\% improvement in PSNR, a 5-6\% increase in SSIM, and upto a 67\% reduction in perceptual error (LPIPS) compared to other transformer approaches.

ProvRain: Rain-Adaptive Denoising and Vehicle Detection via MobileNet-UNet and Faster R-CNN

TL;DR

The paper tackles nighttime vehicle detection under adverse weather by introducing ProvRain, a lightweight denoising-augmented pipeline designed for real-time automotive use.ProvRain integrates a weather-aware MobileNet-U-Net denoiser trained via curriculum learning with a light proposal generator and a Faster R-CNN classifier to robustly detect headlights/taillights and suppress rain-induced noise.Key contributions include the curriculum-based denoising framework, a light-weight perception stack suitable for automotive hardware, and a multi-stage fusion of geometric, temporal, and classification cues that enhances early warning and tracking under rain.Empirical results on the PVDN dataset show improved accuracy, recall, PSNR/SSIM metrics for denoising, and superior early-warning performance compared with baseline methods and heavier transformer-based denoisers.

Abstract

Provident vehicle detection has a lot of scope in the detection of vehicle during night time. The extraction of features other than the headlamps of vehicles allows us to detect oncoming vehicles before they appear directly on the camera. However, it faces multiple issues especially in the field of night vision, where a lot of noise caused due to weather conditions such as rain or snow as well as camera conditions. This paper focuses on creating a pipeline aimed at dealing with such noise while at the same time maintaining the accuracy of provident vehicular detection. The pipeline in this paper, ProvRain, uses a lightweight MobileNet-U-Net architecture tuned to generalize to robust weather conditions by using the concept of curricula training. A mix of synthetic as well as available data from the PVDN dataset is used for this. This pipeline is compared to the base Faster RCNN architecture trained on the PVDN dataset to see how much the addition of a denoising architecture helps increase the detection model's performance in rainy conditions. The system boasts an 8.94\% increase in accuracy and a 10.25\% increase in recall in the detection of vehicles in rainy night time frames. Similarly, the custom MobileNet-U-Net architecture that was trained also shows a 10-15\% improvement in PSNR, a 5-6\% increase in SSIM, and upto a 67\% reduction in perceptual error (LPIPS) compared to other transformer approaches.

Paper Structure

This paper contains 13 sections, 12 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Preprocessing and Weather-Aware Denoising
  5. Light Proposal Generation
  6. Proposal Classification
  7. Pairing of instances
  8. Temporal Tracking
  9. Decision Logic
  10. Data Preprocessing
  11. Observations
  12. Results And Inference
  13. Conclusions and Future work

Figures (5)

  • Figure 1: ProvRain Architecture Diagram
  • Figure 2: Distribution of Light Instances per Vehicle
  • Figure 3: Reflection-to-Direct Light Ratio Across Vehicles
  • Figure 4: Curricula Training Dataset Preperation
  • Figure 5: Detection Performance Comparison