Depth map visualization problem

[ItaloFan]

Thanks for your great work! I get excellent result on my indoor dataset. But when I try to visualize the depth map, I find it's ugly compared with colmap depth map. I just normalize the depth map through (depthmap - depthmap.min)/(depthmap.max - depthmap.min). Is there anything better I can do? colmap depth map: ACMP depth map: Planar model:

[GhiXu]

Hi, @ItaloFan , maybe you can try to use the depth_min and depth depth_max read from XXXXXXXX_cam.txt to exclude some abnormal values, and then use them to normalize depth maps.

[fl8976219]

Thanks for your great work! I used program to read .dmb file by removing the file header and getting a two dimensional array. Then I used plt.imshow to display the dimensional array. My way does not get a good display. Could you tell me how to get the depth map shown above from .dmb files？Thank you very much.

[helloycr]

I can do that. You can find me. My Q: 917318329

[ItaloFan]

Sorry for late reply.

Thanks for your great work! I used program to read .dmb file by removing the file header and getting a two dimensional array. Then I used plt.imshow to display the dimensional array. My way does not get a good display. Could you tell me how to get the depth map shown above from .dmb files？Thank you very much.

Here's my code:

import numpy as np
from PIL import Image

def read_depth(
    input: str, output: str, min_depth: float = None, max_depth: float = None
):
    file = open(input, "rb")
    type_, h, w, nb = np.fromfile(file, dtype=np.int32, count=4)
    depth_map = np.fromfile(file, dtype=np.float32)
    file.close()
    depth_map = np.resize(depth_map, (h, w))
    depth_map[np.isnan(depth_map)] = 0
    if min_depth:
        depth_map[depth_map < min_depth] = min_depth
    if max_depth:
        depth_map[depth_map > max_depth] = max_depth
    if not (min_depth or max_depth):
        min_depth, max_depth = np.quantile(depth_map[depth_map > 0], (0.05, 0.95))
        depth_map[depth_map < min_depth] = min_depth
        depth_map[depth_map > max_depth] = max_depth
    # convert to range [0-1]
    depth_map = (depth_map - depth_map.min()) / (
        depth_map.max() - depth_map.min()
    )
    # pseudo color map
    cm_jet = mpl.cm.get_cmap("jet")
    im = cm_jet(depth_map)
    im = np.uint8(im * (2**8 - 1))
    img = Image.fromarray(im)
    img.save(output) # save picture

By the way, I tried to clip depth using infomation from camera, but it doesn't improve much.