creates tiny imagenet class names from imagenet classnames

create_tiny_imagenet_classnames.py

import os
import json
import requests

def download_file(url, path):
    if not os.path.exists(path):
        response = requests.get(url)
        # if download failed raise exception
        response.raise_for_status()
        with open(path, 'w', encoding='utf-8') as file:
            file.write(response.text)
    with open(path, 'r') as f:
        data = f.read()
    return data

def main():
    imagenet_classes_json_url = 'https://gist.githubusercontent.com/Coderx7/f14bc4c754352f13f9196f64e961f964/raw/22392c8dc1cfa9f5007294d0af1bb8232d7206e7/imagenet_class_index.json'
    imagenet_classes_json_path = './imagenet_class_index.json'

    tinyimagenet_classes_url = 'https://gist.githubusercontent.com/Coderx7/d73fe551fec1bc8edba275c34d6ffb59/raw/211af8bc361682f1f263fe5997442db38788acdf/tinyimagenet_classIds.txt'
    tinyimagenet_classes_path = './tinyimagenet_classIds.txt'
    tinyimagenet_classnames_path = './tinyimagenet_classnames.txt'

    json_data_string = download_file(imagenet_classes_json_url, imagenet_classes_json_path)
    imagenet_data = json.loads(json_data_string)
    imagenet_classname_dict = {}
    for key, value_list in imagenet_data.items():
        if len(value_list) == 2: 
            imagenet_classname_dict[value_list[0]] = value_list[1]
        else:
            print(f"skipping invalid entry in json data for key '{key}': {value_list}")

    tinyimagenet_ids = download_file(tinyimagenet_classes_url, tinyimagenet_classes_path)
    tinyimagenet_ids = [line.strip() for line in tinyimagenet_ids.strip().split('\n')]

    tinyimagenet_classnames = []
    missing_ids = []

    for id in tinyimagenet_ids:
        class_name = imagenet_classname_dict.get(id)
        if class_name is not None:
            tinyimagenet_classnames.append(class_name)
        else:
            print(f"target id '{id}' not found in the provided json data")
            missing_ids.append(id)
            tinyimagenet_classnames.append(f"MISSING_{id}")

    print(f'{len(tinyimagenet_classnames)} labels processed!')
    
    if input('Display all labels?(y/N)').lower() == 'y':
        print(f'Tiny imagenet labels:\n{tinyimagenet_classnames}')

    if missing_ids:
        print("\nERROR: The following target IDs were not found in the JSON data:")
        print(missing_ids)
    else:
        print(f"\nSuccessfully mapped all {len(tinyimagenet_ids)} target ids.")

    assert len(tinyimagenet_classnames) == len(tinyimagenet_ids), "mismatch in list lengths!"

    with open(tinyimagenet_classnames_path, 'w', encoding='utf-8') as f:
        for item in tinyimagenet_classnames:
            f.write(str(item) + '\n')

if __name__ == '__main__':
    main()