街景目标检测数据集4813张VOC+YOLO格式

数据集格式：VOC格式+YOLO格式

压缩包内含：3个文件夹，分别存储图片、xml、txt文件

JPEGImages文件夹中jpg图片总计：4813

Annotations文件夹中xml文件总计：4813

labels文件夹中txt文件总计：4813

标签种类数：26

标签名称:["bicycle","building","bus","car","cctv","crosswalk","curb","guardrail","lanemarking","manhole","motorcycle","obstacle","parking","person","pole","road","road-sign","safety-sign","sidewalk","speed-bump","street-light","terrain","traffic light","traffic-sign","truck","utility-pole"]

标签中文对照：“自行车”、“建筑”、“公共汽车”、“汽车”、“闭路电视”、“人行横道”、“路缘”、“护栏”、“车道标记”、“下水井”、“摩托车”、“障碍物”、“停车场”、“人”、“杆”、“道路”、“路标”、“安全标志”、“人行道”、“减速带”、“路灯”、“地形”、“交通灯”、“车辆标志”、”卡车“电线杆”]

每个标签的框数（注意yolo格式类别顺序不和这个对应，而以labels文件夹classes.txt为准）：

bicycle 框数 = 106

building 框数 = 4898

bus 框数 = 107

car 框数 = 8815

cctv 框数 = 141

crosswalk 框数 = 396

curb 框数 = 241

guardrail 框数 = 612

lanemarking 框数 = 783

manhole 框数 = 886

motorcycle 框数 = 517

obstacle 框数 = 344

parking 框数 = 485

person 框数 = 1952

pole 框数 = 31

road 框数 = 4405

road-sign 框数 = 510

safety-sign 框数 = 453

sidewalk 框数 = 4964

speed-bump 框数 = 586

street-light 框数 = 686

terrain 框数 = 53

traffic light 框数 = 317

traffic-sign 框数 = 40

truck 框数 = 784

utility-pole 框数 = 1287

总框数：34399

图片清晰度（分辨率：像素）：清晰

图片是否增强：是

标签形状：矩形框，用于目标检测识别

重要说明：暂无

特别声明：本数据集不对训练的模型或者权重文件精度作任何保证，数据集只提供准确且合理标注

标注及图片情况如下：