Skip to content

GitLab

Closed
Created by Ruwan Abeywardhana@RCAbeyDeveloper

Few updates are requred to ketos tutorial

following issues were noticed while following tutorial using ketos 2.6

Creating a training database (Basic)

In [7]:
  1. ketos.data_handling.selection_table.standardize() arguments (signal_labels, backgr_labels) have been changed in ketos 2.6
  2. ketos.data_handling.selection_table.standardize() argument 'mapper' key and value are required to swap

Suggestion: In [7]

 map_to_ketos_annot_std ={'filename':'sound_file'} 
 std_annot_train = sl.standardize(table=annot_train, labels=[1], start_labels_at_1=True, 
                             mapper=map_to_ketos_annot_std, trim_table=True)
 std_annot_val = sl.standardize(table=annot_val, labels=[1], start_labels_at_1=True, 
                             mapper=map_to_ketos_annot_std, trim_table=True)

Creating a training database (Extended)

In [8]: (same issue as tutorial 1, In[7])

  1. ketos.data_handling.selection_table.standardize() arguments (signal_labels, backgr_labels) have been changed in ketos 2.6

  2. ketos.data_handling.selection_table.standardize() argument 'mapper' key and value are required to swap

Suggestion: In [8]

 map_to_ketos_annot_std ={'filename':'sound_file'} 
 std_annot_train = sl.standardize(table=annot_train, labels=["upcall"], start_labels_at_1=True,
                             mapper=map_to_ketos_annot_std, trim_table=True)
 std_annot_test = sl.standardize(table=annot_test, labels=["upcall"], start_labels_at_1=True,
                            mapper=map_to_ketos_annot_std, trim_table=True)
In 14:

positives_train = sl.select(annotations=std_annot_train, length=3.0, step=0.5, min_overlap=0.5, center=False) take forever to run. (jupyter notebook crash)

In 19, 21, and 22:

pandas future warning pandas.append

warning message: /.../lib/python3.8/site-packages/ketos/data_handling/selection_table.py:1033: FutureWarning: The frame.append method is deprecated and will be removed from pandas in a future version. Use pandas.concat instead. df = df.append(z, ignore_index=True)

detector.py

line 97
  1. arguments of load_model_file are requred to changed
  2. method load_model_file will be removed in future versions, better to replace with load method

Old: Line 97 model, audio_repr = ResNetInterface.load_model_file(args.model, './narw_tmp_folder', load_audio_repr=True)

Suggestion:

  `model, audio_repr = ResNetInterface.load(model_file='narw.kt', new_model_folder='./narw_tmp_folder', load_audio_repr=True)`
Edited by Ruwan Abeywardhana