import os
import random

image_dir = "litter_detection/negative/images"
target_count = 2000

if not os.path.exists(image_dir):
    print(f"Directory not found: {image_dir}")
    exit(1)

images = [f for f in os.listdir(image_dir) if os.path.isfile(os.path.join(image_dir, f))]
current_count = len(images)

#calculate how many to delete
to_delete = current_count - target_count

#randomly select images to delete
images_to_delete = random.sample(images, to_delete)

# delete images
for i, img in enumerate(images_to_delete, 1):
    filepath = os.path.join(image_dir, img)
    os.remove(filepath)
    if i % 1000 == 0:
        print(f"Deleted {i}/{to_delete}")

print(f"Deleted {to_delete}. Remaining: {target_count}")