Script to generate the labels csv file for AutoML Image Classification (Vertex AI) based on folder structure within google cloud storage

gs_folders_to_labels_file.py

from google.cloud import storage

folder_location = 'YOUR GCS FOLDER LOCATION' # e.g automl-ui-dataset/x-ray-dataset

parts = folder_location.split('/')
bucket_name = parts[0]
prefix = '/'.join(parts[1:]) if len(parts) > 1 else ''

storage_client = storage.Client()
blobs = storage_client.list_blobs(bucket_name, prefix=prefix)

f = open("labels.csv", "w")

for blob in list(blobs)[1:]:
    path = 'gs://' + bucket_name + '/' + blob.name
    label = blob.name.split(prefix)[1].split('/')[1]

    print(path, label)
    f.write(path+','+label+'\n')
f.close()

print(parts, prefix)