meta_data.py

import json
import os
from datasets import Dataset, Features, ClassLabel, Image
from huggingface_hub import login

JSON_FILE = "clouds_1/labels.json"     
DATASET_NAME = "CCAiM-CloudsDataset"         
IMAGE_FOLDER = "clouds_1"        
HUGGINGFACE_TOKEN = input("hugging face aces token: ")
POSSIBLE_SUFFIXES = [".jpg", ".jpeg", ".png", ".bmp", ".webp", "JPG"]

login(token=HUGGINGFACE_TOKEN)

with open(JSON_FILE, "r", encoding="utf-8") as f:
    data_json = json.load(f)

all_labels = sorted(set(data_json.values()))
class_labels = ClassLabel(names=all_labels)

def find_file(prefix, folder):
    for suffix in POSSIBLE_SUFFIXES:
        candidate = os.path.join(folder, prefix + suffix)
        if os.path.exists(candidate):
            return candidate
    raise FileNotFoundError(f"file not found with prefix {prefix} {POSSIBLE_SUFFIXES}")

data_list = []
for prefix, label in data_json.items():
    try:
        image_path = find_file(prefix, IMAGE_FOLDER)
        data_list.append({"image": image_path, "label": label})
    except FileNotFoundError as e:
        print(e)

features = Features({
    "image": Image(),
    "label": ClassLabel(names=all_labels)
})

dataset = Dataset.from_list(data_list, features=features)

dataset.push_to_hub(DATASET_NAME, private=False)
print(f"Dataset {DATASET_NAME} pulled to Hugging Face!")