moondream/car_part_damage

这是一个关于汽车部件损坏的数据集，主要用于图像分割和物体检测任务。数据集包含RGB图像及其标注信息，标注信息包括汽车部件类别、边界框（xyxy坐标）和分割多边形。数据集分为训练集（948张图像）和测试集（50张图像），包含21个汽车部件类别，如前后保险杠、车门、车轮、车窗、挡风玻璃、格栅、头灯、引擎盖、车牌、后视镜、侧板、车顶、尾灯和后备箱等。

This is a dataset about car part damage, primarily used for image segmentation and object detection tasks. The dataset includes RGB images along with annotations such as car part categories, bounding boxes (xyxy coordinates), and segmentation polygons. It is divided into a training set (948 images) and a test set (50 images), covering 21 car part categories including front/rear bumpers, doors, wheels, windows, windshields, grilles, headlights, hoods, license plates, mirrors, quarter panels, rocker panels, roofs, tail lights, and trunks.

搜集汇总

背景与挑战

背景概述

该数据集是一个用于汽车部件损坏检测的图像数据集，包含998张图像，标注了21种汽车部件（如保险杠、车门、车轮等）的边界框和分割多边形。它适用于图像分割和对象检测任务，主要用于训练和评估模型在识别汽车部件损坏方面的性能。