KITTI数据集简析

KITTI数据集

数据集结构

• KITTI数据集网盘 提取码：0bjl

KITTI
├── devkit_object
|   ├── cpp
|   ├── mapping
|   ├── matlab
|   └── readme.txt
│—— data_object_calib
│   ├── testing
│   │   └── calib
│   └── training
│       └── calib
│—— data_object_label_2
│   ├── testing
│       └── label_2
│—— data_object_image_2
│   ├── testing
│   │   └── image_2
│   └──training
│       └── image_2
│—— data_object_velodyne
│   ├── testing
│   │   └── velodyne
│   └── training
│       └── velodyne
│—— data_road
│   ├── testing
│   │   ├── calib
│   │   └── image_2
│   └── training
│       ├── calib
│       ├── gt_image_2
│       └── image_2
│—— depth_2

数据集内容

- data_object_calib 每个样本的标定数据 （测试集7518个，训练集7481个）
- data_object_label_2 3D点云标注文件（只有训练集7481个）
- data_object_image_2 RGB图像 （测试集7518个，训练集7481个）
- data_object_velodyne 激光雷达数据（测试集7518个，训练集7481个）
- data_road 包含测试集和训练集 每集中包含标定参数以及RGB图像 训练集中还多个gt_image_2
- depth_2 7481张深度图像

data_object_calib 样本标定数据

P0：校对后的 camera0 投影矩阵，3x4 数组
P1：校对后的 camera1 投影矩阵，3x4 数组
P2：校对后的 camera2 投影矩阵，3x4 数组
P3：校对后的 camera3 投影矩阵，3x4 数组
R0_rect：校准旋转矩阵，4x4 数组
Tr_velo_to_cam：从 Velodyne 坐标到相机坐标的变换矩阵，4x4 数组
Tr_imu_to_velo：从 IMU 坐标到 Velodyne 坐标的变换矩阵，4x4 数组

data_object_label_2 3D点云标注文件

#Values    Name      Description
----------------------------------------------------------------------------
   1    type         (类别信息:车 货车 卡车...)
                      Describes the type of object: 'Car', 'Van', 'Truck',
                     'Pedestrian', 'Person_sitting', 'Cyclist', 'Tram',
                     'Misc' or 'DontCare'
                     
   1    truncated    (是否被截断:0-1截断程度）
                     Float from 0 (non-truncated) to 1 (truncated), where
                     truncated refers to the object leaving image boundaries
                     
   1    occluded     (是否被遮挡:0完全可见 1部分遮挡 2大部分遮挡 3未知)
                     Integer (0,1,2,3) indicating occlusion state:
                     0 = fully visible, 1 = partly occluded
                     2 = largely occluded, 3 = unknown
                     
   1    alpha        (物体观察角度:[-pi, pi](弧度))
                     Observation angle of object, ranging [-pi..pi]
                     
   4    bbox         (2D图像标注框:左上角右下角坐标)
                     2D bounding box of object in the image (0-based index):
                     contains left, top, right, bottom pixel coordinates
                     
   3    dimensions   (3D点云标注框:高 宽 长(m))
                     3D object dimensions: height, width, length (in meters)
                     
   3    location     (3D点云标注框:相机坐标系底面中心点坐标(m))
                     3D object location x,y,z in camera coordinates (in meters)
                     
   1    rotation_y   (物体正前方向与相机水平方向角度)
                     Rotation ry around Y-axis in camera coordinates [-pi..pi]
                     
   1    score        (得分)
                     Only for results: Float, indicating confidence in
                     detection, needed for p/r curves, higher is better.