upload_to_hf.py

#!/usr/bin/env python3
"""
Upload VR Scene Evaluation Dataset to Hugging Face Hub

This script uploads a YOLO format dataset to Hugging Face Hub as a dataset repository.
For very large datasets, you can also use the command line approach:

1. Install huggingface-hub: pip install huggingface-hub
2. Login: huggingface-cli login
3. Upload: hf upload-large-folder <username>/<repo-name> /path/to/dataset --repo-type=dataset

This Python script provides more control and better error handling.
"""

import os
import yaml
from pathlib import Path
from huggingface_hub import HfApi, login, create_repo
import shutil
import tempfile

# Configuration
DATASET_NAME = "DISCOVR"  # Change this to your desired dataset name
HF_USERNAME = None  # Will be set after login
DATASET_PATH = "/home/daniel/_datasets/post-2/aggregate"
REPO_TYPE = "dataset"

def load_dataset_config():
    """Load the dataset configuration from data.yaml"""
    with open(os.path.join(DATASET_PATH, "data.yaml"), 'r') as f:
        config = yaml.safe_load(f)
    return config

def create_dataset_card(config):
    """Create a README.md file for the dataset"""
    
    class_names = config['names']
    num_classes = config['nc']
    
    readme_content = f"""---
license: cc-by-4.0
task_categories:
- object-detection
language:
- en
tags:
- computer-vision
- object-detection
- yolo
- virtual-reality
- vr
- scene-evaluation
size_categories:
- 1K<n<10K
---

# VR Scene Evaluation Dataset

## Dataset Description

This dataset contains {num_classes} object classes for VR scene evaluation, formatted for YOLO object detection models.

### Classes ({num_classes} total):
{chr(10).join([f"- {i}: {name}" for i, name in enumerate(class_names)])}

## Dataset Structure

```
├── train/
│   ├── images/
│   └── labels/
├── valid/
│   ├── images/
│   └── labels/
├── test/
│   ├── images/
│   └── labels/
└── data.yaml
```

## Usage

### With YOLOv8

```python
from ultralytics import YOLO

# Load a model
model = YOLO('yolov8n.pt')

# Train the model
results = model.train(data='path/to/data.yaml', epochs=100, imgsz=640)
```

### With Hugging Face Datasets

```python
from datasets import load_dataset

dataset = load_dataset("{HF_USERNAME}/{DATASET_NAME}")
```

## License

This dataset is licensed under CC BY 4.0.

## Citation

```
@dataset{{vr_scene_evaluation,
  title={{VR Scene Evaluation Dataset}},
  year={{2025}},
  publisher={{Hugging Face}},
  version={{1.0}},
}}
```

## Original Source

This dataset was originally sourced from Roboflow:
- Workspace: my-workspace-zhz1m
- Project: vr-scene-evaluation-o1hbg
- Version: 6
- URL: https://universe.roboflow.com/my-workspace-zhz1m/vr-scene-evaluation-o1hbg/dataset/6
"""
    return readme_content

def prepare_upload_directory():
    """Prepare a clean directory for upload"""
    upload_dir = tempfile.mkdtemp()
    
    # Copy essential files
    files_to_copy = [
        "data.yaml",
        "README.dataset.txt", 
        "README.roboflow.txt"
    ]
    
    for file in files_to_copy:
        src = os.path.join(DATASET_PATH, file)
        if os.path.exists(src):
            shutil.copy2(src, upload_dir)
    
    # Copy train, valid, test directories
    for split in ["train", "valid", "test"]:
        src_dir = os.path.join(DATASET_PATH, split)
        if os.path.exists(src_dir):
            dst_dir = os.path.join(upload_dir, split)
            shutil.copytree(src_dir, dst_dir)
    
    return upload_dir

def main():
    global HF_USERNAME
    
    print("=== Hugging Face Dataset Upload Script ===")
    print(f"Dataset path: {DATASET_PATH}")
    print(f"Dataset name: {DATASET_NAME}")
    
    # Load dataset config
    try:
        config = load_dataset_config()
        print(f"✓ Loaded dataset config: {config['nc']} classes")
    except Exception as e:
        print(f"✗ Error loading dataset config: {e}")
        return
    
    # Login to Hugging Face
    print("\n1. Logging into Hugging Face...")
    print("You need a Hugging Face account and access token.")
    print("Get your token from: https://huggingface.co/settings/tokens")
    
    try:
        login()
        api = HfApi()
        user_info = api.whoami()
        HF_USERNAME = user_info['name']
        print(f"✓ Logged in as: {HF_USERNAME}")
    except Exception as e:
        print(f"✗ Login failed: {e}")
        print("Make sure you have a valid token and internet connection.")
        return
    
    # Create repository
    repo_id = f"{HF_USERNAME}/{DATASET_NAME}"
    print(f"\n2. Creating repository: {repo_id}")
    
    try:
        create_repo(
            repo_id=repo_id,
            repo_type=REPO_TYPE,
            private=False,  # Set to True if you want a private repo
            exist_ok=True
        )
        print("✓ Repository created/verified")
    except Exception as e:
        print(f"✗ Error creating repository: {e}")
        return
    
    # Prepare upload directory
    print("\n3. Preparing files for upload...")
    try:
        upload_dir = prepare_upload_directory()
        print(f"✓ Files prepared in: {upload_dir}")
        
        # Create README.md
        readme_content = create_dataset_card(config)
        with open(os.path.join(upload_dir, "README.md"), 'w') as f:
            f.write(readme_content)
        print("✓ Dataset card created")
        
    except Exception as e:
        print(f"✗ Error preparing files: {e}")
        return
    
    # Upload to Hugging Face
    print(f"\n4. Uploading to {repo_id}...")
    print("This may take a while depending on dataset size...")
    print("Using upload_large_folder for better handling of large datasets...")
    
    try:
        # Use upload_large_folder for better handling of large datasets
        api.upload_large_folder(
            folder_path=upload_dir,
            repo_id=repo_id,
            repo_type=REPO_TYPE,
            num_workers=4,  # Use multiple workers for faster upload
            create_pr=False,  # Upload directly to main branch
            allow_patterns=["**/*"],  # Upload all files
            ignore_patterns=[".git/**", "**/.DS_Store", "**/__pycache__/**"]  # Ignore system files
        )
        print("✓ Upload completed successfully!")
        print(f"\n🎉 Your dataset is now available at:")
        print(f"https://huggingface.co/datasets/{repo_id}")
        
    except Exception as e:
        print(f"✗ Upload failed: {e}")
        print("If the upload failed due to size, you can try:")
        print("1. Reducing the number of workers (num_workers parameter)")
        print("2. Using the command line: hf upload-large-folder")
        print("3. Splitting the dataset into smaller chunks")
        return
    finally:
        # Clean up
        shutil.rmtree(upload_dir)
        print("✓ Temporary files cleaned up")

if __name__ == "__main__":
    main()