COLLECTION_ID = ''
zc = ZegamiClient("", "")
workspace = zc.get_workspace_by_id(WORKSPACE_ID)
collection = workspace.get_collection_by_id(COLLECTION_ID)
print(collection)
imageset_id = collection._data.get('imageset_id')
resp = nodes.add_node(
zc,
workspace,
'mask_annotation_score',
# truth authors and evaluated authors should either
# be not specified or a list of strings
{
# "evaluated_authors": ["authors name"],
},
'imageset',
imageset_parents=[imageset_id],
name="annotation score ims")
annotation_score_node = resp.get('imageset')
# create mapping dataset to map
join_dataset_id = collection._data.get('imageset_dataset_join_id')
resp = nodes.add_node(zc,
workspace,
'mapping', {},
'dataset',
dataset_parents=[join_dataset_id],
imageset_parents=[annotation_score_node.get('id')],
scaled_imageset_id = collection._data.get('scaled_imageset_id')
join_dataset_id = collection._data.get('imageset_dataset_join_id')
else:
# v2
source = collection.sources[0]
dataset_id = collection._data.get('dataset_id')
scaled_imageset_id = source._data.get('scaled_imageset_id')
join_dataset_id = source._data.get('imageset_dataset_join_id')
resp = nodes.add_node(zc,
workspace,
'custom_feature_extraction', {
"model_name": "feature_extractor.h5",
"greyscale": True,
"width": 164,
"height": 164,
},
'imageset',
imageset_parents=scaled_imageset_id,
name="custom feature extraction node",
node_group=collection_group,
processing_category='image_clustering')
custom_feature_extraction_node = resp.get('imageset')
# Add new clustering node
resp = nodes.add_node(
zc,
workspace,
'cluster',
{
'exclude': [''
# create explainability map node
resp = nodes.add_node(
zc,
workspace,
'explainability_map',
{
"file_name":
MODEL_NAME, # Optional: Name of the model in workspace storage.
# This will overwrite the tensorflow model param.
"model_name":
'Xception', # Optional: Name of the tensorflow model. Default is ResNet50.
"width":
224, # Optional: Width value to pre process the image before inputting into the model.
"height":
224, # Optional: Height value to pre process the image before inputting into the model.
"last_conv_layer_name":
'', # Optional: Name of the last convolutional layer.
"class_index":
'', # Optional: Index of class to evaluate activation maps with respect to.
# If left blank, the top predicted class is used for each image.
"alpha":
0.5, # Optional: Blend weight for combining heatmap to image. Default is 0.4.
},
'imageset',
imageset_parents=augment_imageset_id,
name="explainability map node",
node_group=collection_group,
processing_category='upload')
explainability_map_node = resp.get('imageset')
zc = ZegamiClient()
workspace = zc.get_workspace_by_id(WORKSPACE_ID)
collection = workspace.get_collection_by_id(COLLECTION_ID)
print(collection)
# Add new clustering node
dataset_id = collection._data.get('dataset_id')
scaled_imageset_id = collection._data.get('scaled_imageset_id')
join_dataset_id = collection._data.get('imageset_dataset_join_id')
print('adding feature extraction node')
# create feature extraction node
resp = nodes.add_node(zc,
workspace,
'image_feature_extraction', {},
imageset_parents=[scaled_imageset_id],
type='imageset',
name="Feature extraction imageset")
features_node = resp.get('imageset')
print('\nadded feature extraction node', features_node)
# create clustering node
resp = nodes.add_node(
zc,
workspace,
'cluster', {
"columns_order": [1002, 1003],
"out_column_titles": ["Image Similarity x", "Image Similarity y"],
"out_columns": ["image_similarity_x", "image_similarity_y"]
},
dataset_parents=features_node.get('id'),
WORKSPACE_ID = ''
COLLECTION_ID = ''
zc = ZegamiClient()
workspace = zc.get_workspace_by_id(WORKSPACE_ID)
# import pdb; pdb.set_trace()
collection = workspace.get_collection_by_id(COLLECTION_ID)
print(collection)
augment_imageset_id = collection._data.get('augment_imageset_id')
# create classification imagesets to generate classification data
resp = nodes.add_node(zc,
workspace,
'image_classification', {},
'imageset',
imageset_parents=[augment_imageset_id],
name="classification ims")
classification_node = resp.get('imageset')
# create image_info dataset to render classifications as a tsv file
resp = nodes.add_node(zc,
workspace,
'image_info', {},
'dataset',
imageset_parents=classification_node.get('id'),
name="classification info ds")
info_node = resp.get('dataset')
# create mapping dataset to map
join_dataset_id = collection._data.get('imageset_dataset_join_id')
dataset_id = collection._data.get('dataset_id')
scaled_imageset_id = collection._data.get('scaled_imageset_id')
join_dataset_id = collection._data.get('imageset_dataset_join_id')
else:
# v2
source = collection.sources[0]
dataset_id = collection._data.get('dataset_id')
scaled_imageset_id = source._data.get('scaled_imageset_id')
join_dataset_id = source._data.get('imageset_dataset_join_id')
print('adding feature extraction node')
# create feature extraction node
resp = nodes.add_node(zc,
workspace,
'image_feature_extraction', {},
imageset_parents=[scaled_imageset_id],
type='imageset',
name="Feature extraction " + NAME,
node_group=collection_group,
processing_category='image_clustering')
features_node = resp.get('imageset')
print('\nadded feature extraction node', features_node)
# create clustering node
resp = nodes.add_node(zc,
workspace,
'cluster', {
"columns_order": [1002, 1003],
"out_column_titles": [NAME + " x", NAME + " y"],
"out_columns": [col_name + "x", col_name + "y"]
},
dataset_parents=features_node.get('id'),
COLLECTION_ID = ''
zc = ZegamiClient()
workspace = zc.get_workspace_by_id(WORKSPACE_ID)
collection = workspace.get_collection_by_id(COLLECTION_ID)
print(collection)
# Add new clustering node
dataset_id = collection._data.get('dataset_id')
resp = nodes.add_node(
zc,
workspace,
'cluster',
{
'exclude': [''
], # list of column names as found in tsv files (not Ids)
'out_columns': ['mean_datasimx', 'mean_datasimy'],
'out_column_titles': ['Mean_DataSimX', 'Mean_DataSimY'],
'columns_order': [1100, 1101
] # important to set this to a unique value
},
dataset_parents=dataset_id,
name="Mean data similarity")
cluster_node = resp.get('dataset')
# Include output in collection output
output_dataset_id = collection._data.get('output_dataset_id')
resp = nodes.add_parent(zc, workspace, output_dataset_id,
cluster_node.get('id'))
from zegami_sdk import nodes
from zegami_sdk.client import ZegamiClient
WORKSPACE_ID = ''
COLLECTION_ID = ''
zc = ZegamiClient()
workspace = zc.get_workspace_by_id(WORKSPACE_ID)
collection = workspace.get_collection_by_id(COLLECTION_ID)
print(collection)
imageset_id = collection._data.get('imageset_id')
# create classification imagesets to generate classification data
resp = nodes.add_node(
zc,
workspace,
'mask_annotation',
{
"weights_blob": "mask_rcnn_alloy-scratch-detector_0100.h5",
"model_author_name": "Alloy Scratch Detector_0100",
# TODO add param for num classes
},
'imageset',
imageset_parents=[imageset_id],
name="masks ims"
)
classification_node = resp.get('imageset')