simple_kitti_visualization

https://github.com/zzzxxxttt/simple_kitti_visualization

GitHub - zzzxxxttt/simple_kitti_visualization: simple visualization scripts for Kitti and Semantic Kitti dataset

 

Requirements

• python >= 3.6
• numpy
• matplotlib
• vtk==8.1.2
• mayavi
• skimage
• seaborn
• pyqt5

 

import numpy as np

import seaborn as sns

import mayavi.mlab as mlab

import argparse

parser = argparse.ArgumentParser()

parser.add_argument('--file_id', required=True, default='070070')

args = parser.parse_args()

colors = sns.color_palette('Paired', 5 * 2)

names = ['Car', 'Truck', 'Pedestrian', 'Cyclist', 'Other']

#file_id = '070070'

if __name__ == '__main__':

  # load point clouds

  scan_dir = f'./velodyne/{args.file_id}.bin'

  scan = np.fromfile(scan_dir, dtype=np.float32).reshape(-1, 8)

  scan = scan[:,:4]

  col = scan[:,3]

  # load labels

  label_dir = f'./label_2/{args.file_id}.txt'

 

  with open(label_dir, 'r') as f:

    labels = f.readlines()

     

  with open(f'./calib/{args.file_id}.txt', 'r') as f:

    lines = f.readlines()

    P2 = np.array(lines[2].strip().split(' ')[1:], dtype=np.float32).reshape(3, 4)

    R0 = np.array(lines[4].strip().split(' ')[1:], dtype=np.float32).reshape(3, 3)

    V2C = np.array(lines[5].strip().split(' ')[1:], dtype=np.float32).reshape(3, 4)

  C2V = np.linalg.inv(np.insert(V2C, 3, values=[0, 0, 0, 1], axis=0))[:3,:]

  fig = mlab.figure(bgcolor=(0, 0, 0), size=(1280, 960))

  # draw point cloud

  #plot = mlab.points3d(scan[:, 0], scan[:, 1], scan[:, 2], col, mode="sphere", scale_factor=.05,resolution=100,figure=fig)

  plot = mlab.points3d(scan[:, 0], scan[:, 1], scan[:, 2], scale_factor=.6,resolution=100,figure=fig)

  plot.glyph.scale_mode = 'scale_by_vector'

  plot.mlab_source.dataset.point_data.scalars = col

  for line in labels:

    line = line.split()

    lab, _, _, _, _, _, _, _, h, w, l, x, y, z, rot = line

    h, w, l, x, y, z, rot = map(float, [h, w, l, x, y, z, rot])

    if lab != 'DontCare':

      x_corners = [l / 2, l / 2, -l / 2, -l / 2, l / 2, l / 2, -l / 2, -l / 2]

      y_corners = [0, 0, 0, 0, -h, -h, -h, -h]

      z_corners = [w / 2, -w / 2, -w / 2, w / 2, w / 2, -w / 2, -w / 2, w / 2]

      corners_3d = np.vstack([x_corners, y_corners, z_corners])  # (3, 8)

      # transform the 3d bbox from object coordiante to camera_0 coordinate

      R = np.array([[np.cos(rot), 0, np.sin(rot)],

                    [0, 1, 0],

                    [-np.sin(rot), 0, np.cos(rot)]])

      corners_3d = np.dot(R, corners_3d).T + np.array([x, y, z])

      # transform the 3d bbox from camera_0 coordinate to velodyne coordinate

      #corners_3d = corners_3d[:, [2, 0, 1]] * np.array([[1, -1, -1]])

      corners_3d = np.hstack((corners_3d, np.ones((8, 1), dtype=np.float32)))  # [8, 4]

      corners_3d = np.dot(corners_3d,np.dot(C2V.T, R0.T))

      def draw(p1, p2, front=1):

        mlab.plot3d([p1[0], p2[0]], [p1[1], p2[1]], [p1[2], p2[2]],

                    color=colors[names.index(lab) * 2 + front], tube_radius=None, line_width=4, figure=fig)

      # draw the upper 4 horizontal lines

      draw(corners_3d[0], corners_3d[1], 0)  # front = 0 for the front lines

      draw(corners_3d[1], corners_3d[2])

      draw(corners_3d[2], corners_3d[3])

      draw(corners_3d[3], corners_3d[0])

      # draw the lower 4 horizontal lines

      draw(corners_3d[4], corners_3d[5], 0)

      draw(corners_3d[5], corners_3d[6])

      draw(corners_3d[6], corners_3d[7])

      draw(corners_3d[7], corners_3d[4])

      # draw the 4 vertical lines

      draw(corners_3d[4], corners_3d[0], 0)

      draw(corners_3d[5], corners_3d[1], 0)

      draw(corners_3d[6], corners_3d[2])

      draw(corners_3d[7], corners_3d[3])

  mlab.view(azimuth=100, distance=300, focalpoint=(100,0,40))

  mlab.savefig(filename=f'result/{args.file_id}.png')

  mlab.close()

 

import os

file_lists = os.listdir('./calib')

for i in range(len(file_lists)):

    file_lists[i] = file_lists[i].split('.')[0]

for id in file_lists:

    cmd = os.path.join('python 3dbox_to_cloud.py --file_id ' + id)

    os.system(cmd)