Behavioral segmentation of open field in DeepLabCut, or B-SOID ("B-side"), is a pipeline that pairs unsupervised pattern recognition with supervised classification to achieve fast predictions of behaviors that are not predefined by users.
GNU General Public License v3.0
190
stars
54
forks
source link
error when running BSOID on concatenated output of multiple DLC runs #71
When I concatenate multiple DLC csv's and use the concatenated file as the input to BSOID, I get the following error on the "predict labels and create example videos" step:
Traceback (most recent call last):
File "/home/evicker2/.conda/envs/bsoid/lib/python3.10/runpy.py", line 187, in _run_module_as_main
mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
File "/home/evicker2/.conda/envs/bsoid/lib/python3.10/runpy.py", line 110, in _get_module_details
import(pkg_name)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 1289, in
f_10fps, trained_tsne, scaler, gmm_assignments, classifier, scores = bsoid_build(TRAIN_FOLDERS)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 957, in bsoid_build
f_10fps, trained_tsne, scaler, gmm_assignments, classifier, scores = bsoid_train(train_folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 936, in bsoid_train
filenames, training_data, perc_rect = likelihoodprocessing(train_folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 296, in likelihoodprocessing
filenames, data, perc_rect = import_folders(folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 278, in import_folders
curr_df_filt, perc_rect = adp_filt(curr_df)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 243, in adp_filt
a, b = np.histogram(data_lh[1:, x].astype(np.float))
ValueError: could not convert string to float: 'DLC_resnet50_3056_E235_ALL_singlePlane_May2522May25shuffle1_500000'
Is there something wrong that I'm doing with the formatting of the concatenated csv? Do you have any ideas about what proper formatting would entail in this case, or if the error is totally unrelated to csv formatting?
Hi,
When I concatenate multiple DLC csv's and use the concatenated file as the input to BSOID, I get the following error on the "predict labels and create example videos" step:
Traceback (most recent call last): File "/home/evicker2/.conda/envs/bsoid/lib/python3.10/runpy.py", line 187, in _run_module_as_main mod_name, mod_spec, code = _get_module_details(mod_name, _Error) File "/home/evicker2/.conda/envs/bsoid/lib/python3.10/runpy.py", line 110, in _get_module_details import(pkg_name) File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 1289, in
f_10fps, trained_tsne, scaler, gmm_assignments, classifier, scores = bsoid_build(TRAIN_FOLDERS)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 957, in bsoid_build
f_10fps, trained_tsne, scaler, gmm_assignments, classifier, scores = bsoid_train(train_folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 936, in bsoid_train
filenames, training_data, perc_rect = likelihoodprocessing(train_folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 296, in likelihoodprocessing
filenames, data, perc_rect = import_folders(folders)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 278, in import_folders
curr_df_filt, perc_rect = adp_filt(curr_df)
File "/gpfs/home/evicker2/Github/B-SOID/bsoid_v1_2_1.py", line 243, in adp_filt
a, b = np.histogram(data_lh[1:, x].astype(np.float))
ValueError: could not convert string to float: 'DLC_resnet50_3056_E235_ALL_singlePlane_May2522May25shuffle1_500000'
Is there something wrong that I'm doing with the formatting of the concatenated csv? Do you have any ideas about what proper formatting would entail in this case, or if the error is totally unrelated to csv formatting?
Thanks,
Evan