Explore and label data

Explore and label data#

This notebook uses the napari-easy-augment-batch-dl widget to explore and label the data. If we have a model trained we can also predict using that model.

import napari
viewer = napari.Viewer()

import os
import numpy as np
import napari
from napari_easy_augment_batch_dl import easy_augment_batch_dl

# import the frameworks you want to use.  As part of the import the framework wil be registered
# if it calls the BaseFramework.register_framework() method
try:
    from napari_easy_augment_batch_dl.frameworks.pytorch_semantic_framework import PytorchSemanticFramework
except:
    print('PytorchSemanticFramework not loaded')
try:
    from napari_easy_augment_batch_dl.frameworks.cellpose_instance_framework import CellPoseInstanceFramework
except:
    print('CellPoseInstanceFramework not loaded')
try:
    from napari_easy_augment_batch_dl.frameworks.stardist_instance_framework import StardistInstanceFramework
except:
    print('StardistInstanceFramework not loaded')

raster_geometry not imported.  This is only needed for the ellipsoid rendering in apply_stardist
No module named 'segment_everything'
PytorchSemanticFramework not loaded
CellPoseInstanceFramework not loaded

Label the Image with a Napari widget#

Here we call the napari-easy-augment-batch-dl widget in label_only mode so we can label the images. Optionally if a model has already been trained we can pass it to the Widget. Then we can make predictions and correct them.

batch_dl = easy_augment_batch_dl.NapariEasyAugmentBatchDL(viewer, label_only = False)

viewer.window.add_dock_widget(
    batch_dl
)

data_path = r'../../data'
parent_path = os.path.join(data_path, 'ladybugs1')
model_path = os.path.join(parent_path, 'models')

batch_dl.load_image_directory(parent_path)

model_name = None
model_name = 'stardist_for_ladybugs'
#model_name = 'cellpose_for_ladybugs'


if model_name == 'stardist_for_ladybugs':
    # set the drop down to the StardistInstanceFramework
    batch_dl.network_architecture_drop_down.setCurrentText(StardistInstanceFramework.descriptor)

    # get the stardist widget
    widget = batch_dl.deep_learning_widgets[StardistInstanceFramework.descriptor]

    # load model
    widget.load_model_from_path(os.path.join(model_path, model_name))

elif model_name == 'cellpose_for_ladybugs':
    # set the drop down to the CellPoseInstanceFramework
    batch_dl.network_architecture_drop_down.setCurrentText(CellPoseInstanceFramework.descriptor)
    
    # get the cellpose widget
    widget = batch_dl.deep_learning_widgets[CellPoseInstanceFramework.descriptor]
    
    # load model
    widget.load_model_from_path(os.path.join(model_path, model_name))

    # get the framework object
    framework = batch_dl.deep_learning_project.frameworks[CellPoseInstanceFramework.descriptor]

    # set the parameters for the cellpose framework
    framework.prob_thresh = -1
    framework.flow_thresh = 0.4
    framework.chan_segment = 0 
    framework.chan2 = 1

    # sync parameters on the widget after this we should see the widget in Napari synced with the parameters we set on the framework
    widget.sync_with_framework()
   

CellPoseInstanceFramework not loaded
PytorchSemanticFramework not loaded
No module named 'segment_everything'
Found framework StardistInstanceFramework
Error creating ml labels and features: 'results'
Error creating ml_labels: 'DeepLearningProject' object has no attribute 'ml_labels'
Random Forest ML may not work properly
Adding object boxes layer
Adding predicted object boxes layer
Adding label boxes
Adding object boxes
Adding predicted object boxes
Setting object box classes
Setting predicted object box classes
Loading network weights from 'weights_best.h5'.
Couldn't load thresholds from 'thresholds.json', using default values. (Call 'optimize_thresholds' to change that.)
Using default values: prob_thresh=0.5, nms_thresh=0.4.
Contents