基于open3d对kitti数据集检测结果可视化

前言

KITTI数据集是自动驾驶和计算机视觉领域中一个广泛使用的基准数据集，它提供了丰富的传感器数据，包括激光雷达、相机和GPS等。Open3D是一个功能强大的3D数据处理和可视化库，支持多种3D数据格式。本文将介绍如何使用Open3D对KITTI数据集的检测结果进行可视化。

准备工作

确保你已经安装了Open3D库。如果尚未安装，可以通过pip安装：

pip install open3d

加载点云

    points  = np.fromfile(str(bin_file), dtype=np.float32).reshape(-1, 4)
    mask = points[:,0]>0
    points= points[mask]
    pts = o3d.geometry.PointCloud()    
    pts.points = o3d.utility.Vector3dVector(points[:, :3])

解析检测框

def get_obj_corners(obj):
    center = [obj["Center"]["CenterX"], obj["Center"]["CenterY"], obj["Center"]["CenterZ"]]
    size = [obj["Size"]["ObjectLength"], obj["Size"]["ObjectWidth"], obj["Size"]["ObjectHeight"]]
    yaw = obj["Yaw"]
    rot = np.asmatrix([[math.cos(yaw), -math.sin(yaw)],\
                    [math.sin(yaw),  math.cos(yaw)]])
    plain_pts = np.asmatrix([[0.5 * size[0], 0.5*size[1]],\
                        [0.5 * size[0], -0.5*size[1]],\
                        [-0.5 * size[0], -0.5*size[1]],\
                        [-0.5 * size[0], 0.5*size[1]]])
    tran_pts = np.asarray(rot * plain_pts.transpose())
    tran_pts = tran_pts.transpose()
    corners = np.arange(24).astype(np.float32).reshape(8, 3)
    for i in range(8):
        corners[i][0] = center[0] + tran_pts[i%4][0]
        corners[i][1] = center[1] + tran_pts[i%4][1]
        corners[i][2] = center[2] + (float(i >= 4) - 0.5) * size[2]
    return corners

def render_3dbbox(points, color):
    lines = [[0, 1], [1, 2], [2, 3], [3, 0], 
             [4, 5], [5, 6], [6, 7], [7, 4],
             [0, 4], [1, 5], [2, 6], [3, 7]]
    colors = [color for i in range(len(lines))]
    line_set = o3d.geometry.LineSet()
    line_set.points = o3d.utility.Vector3dVector(points)
    line_set.lines = o3d.utility.Vector2iVector(lines)
    line_set.colors = o3d.utility.Vector3dVector(colors)
    return line_set

结果可视化