Multi-Weather Pothole Detection (MWPD)

The Multi-Weather-Based Pothole Detection Dataset is a comprehensive collection of images designed to aid in developing and evaluating deep learning models for detecting road surface anomalies, particularly potholes, under diverse environmental conditions. Images captured under normal weather, and rainy conditions include variations in lighting, such as daytime, twilight, and nighttime settings. We tried to add high-quality images to ensure the clarity of road surface details. It facilitates the detection of small and partially obscured potholes. In the dataset, potholes are precisely annotated with bounding boxes. This dataset is meticulously curated to reflect weather scenarios, ensuring robust and adaptable pothole detection systems.

多场景天气路面坑洼检测数据集（Multi-Weather-Based Pothole Detection Dataset）是一套综合性图像集合，旨在助力开发与评估用于检测路面异常（尤其是路面坑洼）的深度学习模型，以适配多样环境条件。该数据集的图像采集于正常天气与降雨天气场景，涵盖日间、黄昏、夜间等不同光照变化情况。为保障路面细节的清晰度，我们尽可能纳入高质量图像样本。其可有效支持小型及部分被遮挡的路面坑洼检测任务。数据集中的路面坑洼均通过边界框（bounding boxes）进行精准标注。本数据集经过精心编纂以覆盖各类天气场景，助力打造鲁棒性强、适配性佳的路面坑洼检测系统。

搜集汇总

背景与挑战

背景概述

Multi-Weather Pothole Detection (MWPD)是一个用于坑洼检测的计算机视觉数据集，包含不同天气和光照条件下的道路图像，所有坑洼都经过精确的边界框标注。该数据集由孟加拉国美国国际大学的研究人员整理，采用CC BY 4.0许可协议，总大小为231MB。