---
license: cc-by-nc-sa-4.0
task_categories:
- text-to-image
- image-to-image
language:
- en
tags:
- southpark
- cartoon
- animation
- comedy
- images
- frames
pretty_name: southpark
size_categories:
- 100K<n<1M
---

# South Park

## South Park Images Dataset

***

![South Park.jpg](https://cdn-uploads.huggingface.co/production/uploads/5f57ea2d3f32f12a3c0692e6/_hqev7bG2Aygd2kYvlx_y.jpeg)

***

# Installation

```python
from huggingface_hub import snapshot_download

repo_id = "asigalov61/South-Park"
repo_type = 'dataset'

local_dir = "./South-Park"

snapshot_download(repo_id, repo_type=repo_type, local_dir=local_dir)

```

***

# Make your own dataset

```sh
!pip install opencv-python
```

```python
import cv2
import os
from tqdm import tqdm

#===============================================================================================

def scan_videos(directory, videos_extensions=['.mkv', '.mp4', '.avi']):
    
    video_files = [os.path.join(directory, f) for f in os.listdir(directory) if os.path.splitext(f)[1].lower() in videos_extensions]
    
    return video_files

def extract_frames(video_path, 
                   output_folder, 
                   interval=0.1, 
                   square_size=480, 
                   scale_size=128, 
                   images_ext='.jpg'
                  ):
    
    if not os.path.exists(output_folder):
        os.makedirs(output_folder)
    
    cap = cv2.VideoCapture(video_path)
    fps = cap.get(cv2.CAP_PROP_FPS)
    frame_interval = int(fps * interval)
    frame_count = 0
    total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))

    print('Video file:', os.path.basename(video_path))
    
    with tqdm(total=total_frames, desc='Extracting frames') as pbar:
        
        while True:
            
            ret, frame = cap.read()
            
            if not ret:
                break
                
            if frame_count % frame_interval == 0:
                
                # Calculate the coordinates for cropping the center square
                height, width = frame.shape[:2]
                center_y, center_x = height // 2, width // 2
                half_size = square_size // 2
                top_left_x = max(center_x - half_size, 0)
                top_left_y = max(center_y - half_size, 0)
                bottom_right_x = min(center_x + half_size, width)
                bottom_right_y = min(center_y + half_size, height)
                
                square_frame = frame[top_left_y:bottom_right_y, top_left_x:bottom_right_x]
                
                # Normalize brightness and contrast
                normalized_frame = cv2.normalize(square_frame, None, alpha=0, beta=255, norm_type=cv2.NORM_MINMAX)

                # Resize
                resized_frame = cv2.resize(normalized_frame, (scale_size, scale_size))
                
                frame_name = os.path.join(output_folder, f"frame_{frame_count}{images_ext}")
                cv2.imwrite(frame_name, resized_frame)
                
            frame_count += 1
            
            pbar.update(1)
    
    cap.release()
    
    print(f"Frames extracted to {output_folder}")

#===============================================================================================

videos_dir = 'Videos'
videos_extensions = ['.mkv', '.mp4', '.avi']

frames_output_dir = 'Output'
frames_extraction_interval = 0.1 # FPS * frames_extraction_interval
original_frame_size = 480
final_frame_size = 128
output_frames_extension = '.jpg'

#===============================================================================================

print('=' * 70)
print('Scanning videos dir...')
video_files = scan_videos(videos_dir)
print('Done!')

print('=' * 70)
print('Found', len(video_files), 'video files')
print('=' * 70)

print('Starting extraction...')
print('=' * 70)

for video in video_files:
        
    extract_frames(video, 
                   os.path.join(frames_output_dir, os.path.splitext(os.path.basename(video))[0]),
                   frames_extraction_interval,
                   original_frame_size,
                   final_frame_size,
                   output_frames_extension
                  )

    print('=' * 70)

print('Extraction finished!')
print('=' * 70)

print('Scanning for extracted frames...')

frames_list = list()

for (dirpath, dirnames, filenames) in os.walk(frames_output_dir):
    frames_list += [os.path.join(dirpath, file) for file in filenames if file.endswith(output_frames_extension)]
                    
print('Done!')
print('=' * 70)

print('Found', len(frames_list), 'video frames')
print('=' * 70)

print('Done!')
print('=' * 70)
```

***

### Project Los Angeles
### Tegridy Code 2024