Saved in:
Bibliographic Details
Main Authors: Ogino, Yuka, Shoji, Yuho, Toizumi, Takahiro, Ito, Atsushi
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2411.02799
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909443201957888
author Ogino, Yuka
Shoji, Yuho
Toizumi, Takahiro
Ito, Atsushi
author_facet Ogino, Yuka
Shoji, Yuho
Toizumi, Takahiro
Ito, Atsushi
contents We propose an image-adaptive object detection method for adverse weather conditions such as fog and low-light. Our framework employs differentiable preprocessing filters to perform image enhancement suitable for later-stage object detections. Our framework introduces two differentiable filters: a Bézier curve-based pixel-wise (BPW) filter and a kernel-based local (KBL) filter. These filters unify the functions of classical image processing filters and improve performance of object detection. We also propose a domain-agnostic data augmentation strategy using the BPW filter. Our method does not require data-specific customization of the filter combinations, parameter ranges, and data augmentation. We evaluate our proposed approach, called Enhanced Robustness by Unified Image Processing (ERUP)-YOLO, by applying it to the YOLOv3 detector. Experiments on adverse weather datasets demonstrate that our proposed filters match or exceed the expressiveness of conventional methods and our ERUP-YOLO achieved superior performance in a wide range of adverse weather conditions, including fog and low-light conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2411_02799
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle ERUP-YOLO: Enhancing Object Detection Robustness for Adverse Weather Condition by Unified Image-Adaptive Processing
Ogino, Yuka
Shoji, Yuho
Toizumi, Takahiro
Ito, Atsushi
Computer Vision and Pattern Recognition
We propose an image-adaptive object detection method for adverse weather conditions such as fog and low-light. Our framework employs differentiable preprocessing filters to perform image enhancement suitable for later-stage object detections. Our framework introduces two differentiable filters: a Bézier curve-based pixel-wise (BPW) filter and a kernel-based local (KBL) filter. These filters unify the functions of classical image processing filters and improve performance of object detection. We also propose a domain-agnostic data augmentation strategy using the BPW filter. Our method does not require data-specific customization of the filter combinations, parameter ranges, and data augmentation. We evaluate our proposed approach, called Enhanced Robustness by Unified Image Processing (ERUP)-YOLO, by applying it to the YOLOv3 detector. Experiments on adverse weather datasets demonstrate that our proposed filters match or exceed the expressiveness of conventional methods and our ERUP-YOLO achieved superior performance in a wide range of adverse weather conditions, including fog and low-light conditions.
title ERUP-YOLO: Enhancing Object Detection Robustness for Adverse Weather Condition by Unified Image-Adaptive Processing
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2411.02799