cogsys-tuebingen / cbmot Goto Github PK

In tracking.py: self.loaded_model = torch.load(model_path) but the model is not initialized anywhere before that.
How to use your pretained model?

Thanks!

By reading your paper, I think score updating can be achieved through simple functions, but why did you choose to use in-depth learning?
And I don't understand what role it plays in the score update function.
Can you help me answer my question?

Hi, thanks for your interesting work !

I'm wondering if there is a plan for releasing the code of your work?

Thanks!

Thanks for sharing your work!
I can't find your launch script "results.sh"!
Is it possible to use the nn learning_score_update_function with a tracker without fusion (only with lidar data)? can you please guide how to use it?

Hello I want to ask what should the file structure looks like?
I don't fully understand the hyperlink between Nuscenes_data_set and ./nuScenes
I download the Nuscenes dataset like below

Then how should I do to link the data

Now my file structure looks like
CBMOT
|--data
|----sets
|-------Nuscenes
|---------samples
|---------sweeps
|---------maps
|---------v1.0-*

Could you help me to figure out how does the file structure should be?
Thanks!!!

Hi,
I want to visulize the tracking results with tracking IDs. Do you have the relevant code? If not, can you tell me how to do it?
I'm looking forward to your reply

Command :

python main.py --work_dir work_dir2 --max_age 3 --Lidar_traindata True --hungarian True

Error message:

Traceback (most recent call last):
  File "main.py", line 362, in <module>
    eval_tracking()
  File "main.py", line 215, in main

  File "/home/user/CBMOT/tracker.py", line 325, in step_centertrack
    matching = comparing_positions(self, self.tracks, results, tracks, dets)
  File "/home/user/CBMOT/tracker.py", line 73, in comparing_positions
    unmatched_positions1_data = [d for d in range(positions1.shape[0]) if not (d in matched_indices[:, 1])]
  File "/home/user/CBMOT/tracker.py", line 73, in <listcomp>
    unmatched_positions1_data = [d for d in range(positions1.shape[0]) if not (d in matched_indices[:, 1])]
TypeError: tuple indices must be integers or slices, not tuple

Thanks for sharing your research. I'm trying to reproduce the results of "Results.sh" from the scratch and have a question about the Centerinput file of "work_dir4".
How to generate the "centertrack_tracks.json" from the output of CenterTrack?
Could I make the json with CenterTrack's nuScenes 3D tracking model? According to your paper, it is necessary to re-managing tracking IDs. Would you please tell me some more details?

Any plans to make the code publicly available?