ms-predict.py

import os
import requests
import percache

iteration = '<YOUR ITERATION ID HERE>'
confidance = 0.999

cache = percache.Cache('ms-predict-cache-{}'.format(iteration))

base_url = 'https://westus2.api.cognitive.microsoft.com/customvision/v3.0/Prediction/<YOUR PROJECT ID HERE>/classify/iterations/{}/image'.format(iteration)

headers = {
    'Prediction-Key': '<YOUR PREDICTION KEY HERE>',
    }


@cache
def classify(url, filename):
    files = {
        'file': (f, open(filename, 'rb'))
    }
    response = requests.post(url, headers=headers, files=files).json()
    return response


@cache
def get_tags(filename):
    list_tags_cmd = '{} -N -l {}'.format(tag_cmd, filename)
    tags = []
    rv = os.popen(list_tags_cmd).read().strip()
    if rv != '':
        tags = rv.split(',')
    return tags


print('Starting using {}'.format(iteration))

dir_path = 'fetch-kats/thumbnails/'
tag_cmd = '~/brew/bin/tag'
for f in os.listdir(dir_path):
    if 'chroniclingamerica.loc.gov' in f:
        continue
    filename = dir_path + f

    # 'fetch-kats/thumbnails/newspapers.com-1918-02-03-page-48-id-457845669.jpeg'
    tags = get_tags(filename)
    if len(tags) > 0:
        print('.', end='', flush=True)
        continue

    response = classify(base_url, filename)
    predictions = {}
    if 'predictions' not in response:
        continue
    for prediction in response['predictions']:
        predictions[prediction['tagName']] = prediction['probability']
    krazy = predictions['krazy']
    if krazy < confidance:
        print(',', end='', flush=True)
        continue

    cmd = '{} -a "Purple" {}'.format(tag_cmd, filename)
    os.system(cmd)
    print("\n{}: {}".format(f, krazy))