Wrong MultiviewX grid visualization

[JohnPekl]

I use calibration from MultivewX to run grid_visualize.py, code shown below.

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import cv2
from multiview_detector.utils import projection
from multiview_detector.datasets.MultiviewX import MultiviewX

if __name__ == '__main__':
    img = Image.open('D:/dataset/tracking/CMC/data/images/MultiviewX/Image_subsets/C1/0000.png')
    dataset = MultiviewX('D:/dataset/tracking/CMC/data/images/MultiviewX')
    xi = np.arange(0, 480, 40)
    yi = np.arange(0, 1440, 40)
    world_grid = np.stack(np.meshgrid(xi, yi, indexing='ij')).reshape([2, -1])
    world_coord = dataset.get_worldcoord_from_worldgrid(world_grid)
    img_coord = projection.get_imagecoord_from_worldcoord(world_coord, dataset.intrinsic_matrices[0],
                                                          dataset.extrinsic_matrices[0])
    img_coord = img_coord[:, np.where((img_coord[0] > 0) & (img_coord[1] > 0) &
                                      (img_coord[0] < 1920) & (img_coord[1] < 1080))[0]]
    plt.imshow(img)
    plt.show()
    img_coord = img_coord.astype(int).transpose()
    img = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
    for point in img_coord:
        cv2.circle(img, tuple(point.astype(int)), 5, (0, 255, 0), -1)
    img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    img.save('img_grid_visualize_mx.png')
    plt.imshow(img)
    plt.show()
    pass

However, the result of grid visualization seems not to be corrected. Could you please help me correct it?

Here is the grid visualization for the Wildtrack dataset

[hou-yz]

please check these two lines in your code, as the current version refers to a ground plane shape of [480, 1440], which only matches the Wildtrack dataset but not the MultiviewX dataset. try [640, 1000] instead.

    xi = np.arange(0, 480, 40)
    yi = np.arange(0, 1440, 40)