baby.seg_trainer.SegFilterParamOptim¶
- class baby.seg_trainer.SegFilterParamOptim(flattener, basic_params={}, IoU_thresh=0.5, scoring='F0_5', nbootstraps=10, bootstrap_frac=0.9)¶
Bases:
object
- # TODO What does this class do
What are the parameters and what do they mean
What are the defaults, what are the ranges/admissible options?
- Parameters
flattener –
basic_params –
IoU_thresh –
scoring –
nbootstraps –
bootstrap_frac –
- Attributes
- basic_params
- opt_params
- opt_score
scoring
The scoring method used during evaluation of the segmentation.
- segrps
- stat_table
- stat_table_bootstraps
- truth
- truth_bootstraps
Methods
filter_trial
fit_filter_params
generate_stat_table
- __init__(flattener, basic_params={}, IoU_thresh=0.5, scoring='F0_5', nbootstraps=10, bootstrap_frac=0.9)¶
Methods
__init__
(flattener[, basic_params, ...])filter_trial
(pedge_thresh, ...[, bootstrap, ...])fit_filter_params
([lazy, bootstrap])generate_stat_table
(example_gen)Attributes
basic_params
opt_params
opt_score
The scoring method used during evaluation of the segmentation.
segrps
stat_table
stat_table_bootstraps
truth
truth_bootstraps
- property scoring¶
The scoring method used during evaluation of the segmentation. Accepted values are: # TODO define the scoring metrics * precision: * recall: * F1: * F0_5: * F2: * meanIoU: :return: str scoring method