randomly choose 500 pictures from food dataset

download_dataset.py

@@ -1,25 +1,25 @@
 import os
 from datasets import load_dataset
 from PIL import Image
+import random
 
 def download_images(name="food101"):
-    # Load the "food101" dataset
-    dataset = load_dataset(name, split="train[:1%]")  # Get a small percentage of the data
+    dataset = load_dataset(name, split="train")
 
     # Create a directory to save the images if it doesn't exist
     output_dir = "data/pictures"
     os.makedirs(output_dir, exist_ok=True)
 
-    # Limit to 200 images
-    num_images = 200
+    # Limit to 500 images
+    num_images = 500
     count = 0
 
-    # Iterate over the dataset and save the images
-    for example in dataset:
-        if count >= num_images:
-            break
+    # Randomly select and save the images
+    random_indices = random.sample(range(len(dataset)), num_images)
+    for index in random_indices:
+        example = dataset[index]
         image = example['image']
         image.save(os.path.join(output_dir, f"image_{count}.jpg"))  # Save as JPG
         count += 1
 
-    print(f"Downloaded and saved {count} images to the folder '{output_dir}'")
+    print(f"Downloaded and saved {num_images} images to the folder '{output_dir}'")