load_json.py

NUM_CLASSES = 228

with open("data/train.json") as train:
  train_json = json.load(train)
  
def generate_label_array(json_obj):
    result = []
    for data in json_obj['annotations']:
        temp_array = [0] * NUM_CLASSES
        for elem in data['labelId']:
            temp_array[int(elem) - 1] = 1
        result.append(temp_array)
    return np.array(result)
    
train_urls = [obj['url'] for obj in train_json['images']]
train_labels = generate_label_arrays(train_json)

print(train_urls[0])
# => https://contestimg.wish.com/api/webimage/570f35feb2f4b95d223aa9b1-large