Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

when i start training,some error notice. #395

Open
gxkjj opened this issue Oct 8, 2024 · 8 comments
Open

when i start training,some error notice. #395

gxkjj opened this issue Oct 8, 2024 · 8 comments

Comments

@gxkjj
Copy link

gxkjj commented Oct 8, 2024

when i start training,some error notice.

SIMBA INVALID ML TRAINING SET ERROR: 11 feature columns exist in some files, but missing in others. The feature files are found in the project_folder/csv/targets_inserted directory. SimBA expects all files within the project_folder/csv/targets_inserted directory to have the same number of features: the first 10 column names with mismatches are: ['Mouse_nose_to_tail', 'Mouse_Ear_distance', 'Movement_mouse_nose', 'Movement_mouse_tail_base', 'Movement_mouse_left_ear', 'Movement_mouse_right_ear', 'M1_largest_euclidean_distance_hull', 'M1_smallest_euclidean_distance_hull', 'M1_mean_euclidean_distance_hull']

@sronilsson
Copy link
Collaborator

Hi @gxkjj, this error happens when SimBA opens all of your files inside the project_folder/csv/targets_inserted directory and concatenate all of those files together.

In your case, when SimBA tries to do this, it finds that those files inside that directory do not have the same column fields. In your case, there are 11 columns that does not exist in all of the files inside the project_folder/csv/targets_inserted directory, and the names of 10 of them are printed in the error message. Do you know if there are any odd CSV files in this directory, or have an inkling about why some file(s) maybe different than others?

@gxkjj
Copy link
Author

gxkjj commented Oct 9, 2024

@sronilsson Thank you for your reply.
There is only one csv file in this path,project_folder/csv/targets_inserted ,this has been bugging me for days, including when I rebuilt the project, and it still does, and I've checked the dlc's files as well, and I don't see these.
TEST2.csv This is my csv file.

@sronilsson
Copy link
Collaborator

sronilsson commented Oct 9, 2024

@gxkjj Got it, three questions:

(i) When you open the project_folder/logs/video_info.csv, what does it say for video TEST2 under the pixels per millimeter heading, and what does it say under the fps heading?

(ii) Which version are you running for simba, i.e., which version do you see if you type pip show simba-uw-tf-dev ?

(iii) Can you send me a full screenshot of the error message in the operating system terminal?

@gxkjj
Copy link
Author

gxkjj commented Oct 10, 2024

@sronilsson Thank you for your reply,
(i)
Screenshot from 2024-10-10 10-32-40

(ii)
Screenshot from 2024-10-10 10-34-55

(iii)
Screenshot from 2024-10-10 10-31-01
Above is a screenshot of my correspondence to the 3 questions. Thank you very much.

@gxkjj
Copy link
Author

gxkjj commented Nov 11, 2024

@sronilsson Can you help me with this, thank you very much!

@sronilsson
Copy link
Collaborator

Hi @gxkjj - I'm sorry I let this slip, thanks for the nudge.

In your project_folder/logs/video_info.csv I see a pixel per millimeter convertion factor that is 0:

image

That is causing the problem.

Can you try the following:

  1. Set the pixel millimeter convertion factor to a value above 0. Either manually in the file, or in the VIDEO PARAMETERS tab.
  2. Extract your features again (new files will be created in the project_folder/csv/features_extracted directory) with the correct pixel per millimeter conversion factor..
  3. Label your behaviors again (OR append the labels to your new files in the project_folder/csv/features_extracted folder, and place the new files in the project_folder/csv/targets_inserted folder.
    4). Train your model again and let me know how it goes.

@gxkjj
Copy link
Author

gxkjj commented Nov 12, 2024

@sronilsson Thank you very much. It was a successful training. Thank you!

@sronilsson
Copy link
Collaborator

Thanks @gxkjj - I will insert some warning message in simba, when it finds pixel per milliliter value of 0.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants