Using Other models/Results
Karobben opened this issue · comments
Dear developer,
I just found that your model is a little bit overfit your data. But we have no experience to train our model.
Would it be possible to use the model from other projects?
I just found the model from Cellpose is very robust and powerful.
I also tried to use the tracking software to rung the result from the Cellpose since the mask result is similar. I loaded it but there are error codes:
2021-12-10 17:19:59,594 ================= Computing cell tracks =================
Traceback (most recent call last):
File "cell_main.py", line 112, in cell_tracking_clicked
trj, col_tuple, col_weights = track_cells(path_in, merged_masks, min_cell_id=1)
File "/mnt/8A26661926660713/Github/Usiigaci/Usiigaci/Tracker/cell_tracking.py", line 20, in track_cells
trj = calculate_initial_cell_info(raw_masks, n_active_features, idx_start_active_features,
File "/mnt/8A26661926660713/Github/Usiigaci/Usiigaci/Tracker/cell_tracking.py", line 93, in calculate_initial_cell_info
trj = tp.link_df(features,
File "/mnt/8A26661926660713/Github/Usiigaci/tensorflow/lib/python3.8/site-packages/trackpy/linking.py", line 559, in link_df
if (is_pandas_since_016 and features.is_copy is not None and
File "/home/ken/.local/lib/python3.8/site-packages/pandas/core/generic.py", line 5487, in __getattr__
return object.__getattribute__(self, name)
AttributeError: 'DataFrame' object has no attribute 'is_copy'
So, would it be possible to use the model form cellpose directly? Or there is a way to run the results from cellpose to processing tracking? It would be very convenient and helpful.
Thank you