Quick Jupyter notebook to auto-tag Magic card artwork from Scryfall using Resnet

ScryfallAutoTagger.ipynb

# Load our classifier using Keras and pre-load wieghts trained from ImageNet

# c.f. https://github.com/Graystripe17/ResNet50-Demo

from keras.applications.resnet50 import ResNet50
from keras.preprocessing import image
from keras.applications.resnet50 import preprocess_input, decode_predictions
import numpy as np
import glob
from IPython.display import Image, display

model = ResNet50(weights='imagenet')

def predict(img_path):
    img = image.load_img(img_path, target_size=(224, 224))
    x = image.img_to_array(img)
    x = np.expand_dims(x, axis=0)
    x = preprocess_input(x)
    preds = model.predict(x)
    decoded = decode_predictions(preds, top=10)[0];
    # print('Predicted:', decoded)
    return decoded

# Load ScryFall bulk data (note, must manually download this)
import json
import urllib.request

data = []

# Download bulk data if we don't already have a local copy
urllib.request.urlretrieve('https://archive.scryfall.com/json/scryfall-default-cards.json', 'scryfall-default-cards.json')
with open('scryfall-default-cards.json') as json_file:
    data = json.load(json_file)

# Loop throuch every card
for c in data:
    # Grab the art_crop of every card (if it exists)...
    if 'image_uris' in c:
        if 'art_crop' in c['image_uris']:
            # Cache it locally
            localpath = "art_crop/" + c['id'] + ".jpg"
            urllib.request.urlretrieve(c['image_uris']['art_crop'], localpath)
            
            # Run predictions on the image
            preds = predict(localpath)
            for pred in preds:
                # Filter out anything < 40% confidence
                if pred[2] > 0.4:
                    # HACK: Manually filter out "Comic Book" because that's a common label for this style of artwork
                    if pred[1] != 'comic_book': 
                        # If we have a valid prediction, then display the image and output the data for it.
                        display(Image(filename=localpath))
                        print(pred[1], pred[2], c['name'], c['id'])