洪灾图文基础数据集(FloodMulS、FloodIT)
收藏资源简介:
该数据集字段包括,参照物类型:car(0)、license plate(1)、motorcycle(2)、person(3)、water(4)、wheel(5),分割的像素点坐标(每行连续的小数中,连续两个表示某个像素点在横/纵向的相对位置);使用“训练得到的多参照物检测模型”生成的参照物边界框信息(Bounding Box)和分割的像素点坐标皆按图片名存储在对应的txt文件中,图片统一是JPG格式。
This dataset includes the following fields: reference object types: car (0), license plate (1), motorcycle (2), person (3), water (4), wheel (5); segmented pixel coordinates (each line contains consecutive decimal numbers, where every two consecutive decimals represent the horizontal and vertical relative position of a pixel); both the bounding box information of reference objects generated by the "trained multi-reference object detection model" and the segmented pixel coordinates are stored in corresponding TXT files sorted by image names, and all images are uniformly in JPG format.




