maryzhang/hw1-24679-image-dataset

Asian vs Western Food Classification Dataset

Dataset Summary

Purpose: This dataset was created for binary classification of food images into Asian or Western cuisine categories, developed as part of CMU 24-679 coursework to explore computer vision techniques in food recognition.

Quick Stats:

• 360 total images (40 original + 320 augmented)
• Binary classification task
• 224x224 RGB images
• Balanced classes (~50% each category)

Contact: maryzhang@cmu.edu

Sample Contact Sheet

Sample grid showing 4 Asian cuisine images (top row) and 4 Western cuisine images (bottom row) from the original dataset

Dataset Composition

Features

• image: PIL Image object (224x224 RGB)
• label: Integer (0=Western, 1=Asian)

Class Distribution

Cuisine Type	Original	Augmented	Label
Western	20	160	0
Asian	20	160	1

Data Splits

• original: 40 hand-collected food images
• augmented: 320 synthetically augmented images (8x augmentation per original)

Data Collection Process

Collection Methodology

Images were collected between January-February 2025 using:

• Personal photography of restaurant meals
• Home-cooked dishes from both culinary traditions
• Mobile phone cameras (various models)
• Natural lighting conditions when possible

Selection Criteria

• Clear food presentation
• Distinctive cuisine characteristics
• No people or identifying information in frame
• Variety in dish types (appetizers, mains, desserts)
• Representative of common dishes from each cuisine

Preprocessing and Augmentation

Preprocessing Pipeline

1. Resize to 224x224 pixels (bilinear interpolation)
2. Convert to RGB format
3. Normalize pixel values to [0, 255]

Augmentation Techniques

Each original image generated 7 augmented variants using:

• Geometric: RandomResizedCrop (0.7-1.0), RandomRotation (±15°), RandomHorizontalFlip (p=0.5)
• Color: ColorJitter (brightness=0.2, contrast=0.2, saturation=0.15, hue=0.05)
• Advanced: RandAugment (num_ops=2, magnitude=7), RandomErasing (p=0.2)

Labels and Annotation

Labeling Schema

• 0: Western cuisine (European, American dishes)
• 1: Asian cuisine (East Asian, Southeast Asian dishes)

Annotation Process

• Manual labeling by dataset creator
• Verification based on known cuisine origins
• Edge cases resolved by primary ingredients and cooking methods

Intended Use and Limitations

Intended Use Cases

• Educational projects in computer vision
• Binary food classification research
• Transfer learning experiments
• Baseline model development

Limitations

• Small dataset size requires transfer learning for good performance
• Binary classification oversimplifies cuisine diversity
• May not generalize to fusion cuisine or ambiguous dishes
• Limited to common dishes, not comprehensive of all cuisine varieties
• Potential bias toward restaurant-style presentation

Out-of-Scope Uses

• Commercial food recognition systems
• Medical or dietary assessment applications
• Fine-grained cuisine classification (regional variations)
• Production deployment without additional training data

Ethical Considerations

Representation

• Effort made to balance representation between cuisine types
• Acknowledges that binary classification may perpetuate oversimplified cultural categories
• "Asian" and "Western" are broad categorizations that don't capture cuisine diversity

Privacy

• No personally identifiable information included
• No restaurant branding or identifying markers
• Images taken in public settings or personal kitchens

Cultural Sensitivity

• Dataset created for educational purposes
• Users should be aware of cultural simplification in binary categorization
• Not intended to make value judgments about cuisine types

AI Usage Disclosure

AI-Assisted Components

• Augmentation code: Partially generated using AI assistance for transform pipeline
• Documentation: README structure and sections refined with AI assistance
• Data collection: All images are original, not AI-generated
• Labels: Manually assigned, not AI-generated

Human Oversight

• All images personally collected and verified
• Labels manually reviewed for accuracy
• Augmentation parameters tuned based on empirical testing
• Final dataset curated and quality-checked by human creator

Usage Example

from datasets import load_dataset
from torchvision import transforms

# Load dataset
dataset = load_dataset("maryzhang/hw1-24679-image-dataset")

# Setup transforms
transform = transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
])

# Access data
sample = dataset['original'][0]
image, label = sample['image'], sample['label']
cuisine = "Asian" if label == 1 else "Western"

print(f"Sample cuisine type: {cuisine}")

Citation

@dataset{zhang2025food,
  author = {Mary Zhang},
  title = {Asian vs Western Food Classification Dataset},
  year = {2025},
  publisher = {Hugging Face},
  note = {CMU 24-679 Homework 1},
  url = {https://huggingface.co/datasets/maryzhang/hw1-24679-image-dataset}
}

License

This dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Contact

Dataset created by Mary Zhang for CMU 24-679. For questions or issues, please use the discussion forum on Hugging Face.