UZ-SLAMLab / Endo-Depth-and-Motion

GNU General Public License v3.0
95 stars 15 forks source link

about tracking #29

Closed 309020726 closed 1 year ago

309020726 commented 1 year ago

python apps/tracking_ours/main.py -d cuda:0 -i path/to/hamlyn_tracking_test_data/test1 -o apps/tracking_ours/results [Open3D WARNING] GLFW Error: X11: The DISPLAY environment variable is missing [Open3D WARNING] Failed to initialize GLFW [Open3D WARNING] GLFW Error: X11: The DISPLAY environment variable is missing [Open3D WARNING] Failed to initialize GLFW [I 231011 16:39:40 main:125] Reseting scene. Segmentation error (core dumped)

how to deal with this problem? Thanks for your help.

DavidRecasens commented 1 year ago

Hi,

It is related to Open3D and so it needs a display/screen connected to the machine where you are executing the code. If you are trying to execute it in a screenless machine like a remote cluster, maybe you can try with "xvfb":

xvfb-run python apps/tracking_ours/main.py -d cuda:0 -i path/to/hamlyn_tracking_test_data/test1 -o apps/tracking_ours/results

But I have not tried this. If not, you can try modifying a bit the code to not execute the visualization part, which is what uses Open3D.

Hope this helps :).

309020726 commented 1 year ago

It‘s ok, thanks for your help.

309020726 commented 1 year ago

Hello, here's another question.How do I fix this?

[I 231016 19:37:53 main:125] Reseting scene. [I 231016 19:37:58 main:186] KEYFRAME INSERTED Computing time 5065ms Traceback (most recent call last): File "apps/tracking_ours/main.py", line 424, in main() File "apps/tracking_ours/main.py", line 240, in main (error_np, synthetic_image_np) = synthetise_image_and_error( File "/homeu/disk2/lxj/Endo-Depth-and-Motion-main/edam/optimization/utils_frame.py", line 35, in synthetise_image_and_error i_points_3d_i = _backproject_points( File "/homeu/disk2/lxj/Endo-Depth-and-Motion-main/edam/geometry/depth_to_flow.py", line 23, in _backproject_points i_points_3d_i: torch.Tensor = kornia.depth_to_3d( TypeError: depth_to_3d() got an unexpected keyword argument 'normalize_points'

DavidRecasens commented 1 year ago

Hi,

Make sure that you are using a version of Kornia where the function "kornia.depth_to_3d" has the argument "normalize_points", like kornia 0.6.0.

309020726 commented 1 year ago

Your help was very much appreciated.

kaikoko commented 1 year ago

Can someone help me deal with this problem? Thanks for your help.

Traceback (most recent call last): File "apps/tracking_ours/main.py", line 427, in main() File "apps/tracking_ours/main.py", line 236, in main pe.run(new_frame, True) File "/home/liu/disk2/nxq/cook/Endo-Depth-and-Motion-main/edam/optimization/pose_estimation.py", line 146, in run gauss_newton( File "/home/liu/disk2/nxq/cook/Endo-Depth-and-Motion-main/edam/optimization/optimization_algorithms.py", line 41, in gauss_newton torch.pinverse(jac[torch.abs(error_o) > 0, :]) RuntimeError: svd_cuda: For batch 0: U(4,4) is zero, singular U.

cho-wang001 commented 5 months ago

python apps/tracking_ours/main.py -d cuda:0 -i path/to/hamlyn_tracking_test_data/test1 -o apps/tracking_ours/results [Open3D WARNING] GLFW Error: X11: The DISPLAY environment variable is missing [Open3D WARNING] Failed to initialize GLFW [Open3D WARNING] GLFW Error: X11: The DISPLAY environment variable is missing [Open3D WARNING] Failed to initialize GLFW [I 231011 16:39:40 main:125] Reseting scene. Segmentation error (core dumped)

how to deal with this problem? Thanks for your help.

Hi,

Could you implement correct tracking without visualizations after changing the code? I modified the code so that it can be traced without visualization. But the pose estimation was incorrect. Every two consecutive frames have the same pose.

Hope you have any thought on this.

Thanks!