Dark Mode
Google Cartoon Face Metadata Extension
Product Reviews & Feedback
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
Provides tabulated attribute information for a subset of 100,000 cartoon faces generated from the Google Cartoon Set. This dataset consolidates complex metadata, originally scattered across many individual files, into easily digestible CSV tables. Its primary purpose is to broaden the usability of the image set for researchers engaging in sophisticated tasks like classifier modelling and various forms of categorical generative modelling, such as CGANs. It allows users to quickly correlate visual features with structured data tags.
Columns
The data product includes two distinct files:
- cartoon_image_attributes.csv: Contains rows corresponding to an image within the 100K set. Columns list the image's filename alongside its 18 specified attributes (e.g., eye_color, hair_type).
- cartoon_attribute_variants.csv: Contains 2 columns detailing the attributes and their variance.
attribute: Defines the specific face characteristic (e.g., eye_angle).variants: Notes the total number of possible options or variants for that particular attribute.
Distribution
The data is supplied in CSV format. The two primary files are designed to accompany the existing 100K JPEG image set. The
cartoon_attribute_variants.csv file has 2 columns and 18 rows, reflecting the 18 unique attributes tracked. The main attribute file, cartoon_image_attributes.csv, contains metadata linked to the 100,000 image records.Usage
- Classifier Modelling: Training models to predict or identify specific attributes within cartoon images.
- Categorical Generative Modelling (e.g., CGANs): Creating new, targeted images based on desired attribute combinations.
- Image Analysis Tasks: Performing statistical analysis on visual data structure and feature frequency.
Coverage
This dataset focuses entirely on synthetically generated cartoon faces. The scope is defined by the 18 specific facial and stylistic attributes used in the generation process. It does not relate to human demographics, geography, or time ranges. The attribute coverage includes elements like 111 different hair type options and 5 eye colour options.
License
CC BY-SA 4.0
Who Can Use It
- Machine Learning Developers: Building and testing generative models (CGANs).
- Computer Vision Experts: Developing feature extraction techniques and image classification algorithms.
- Researchers: Studying attribute correlations and complex visual patterns in large-scale generated datasets.
Dataset Name Suggestions
- Cartoon Set 100K Attribute Mapping
- Google Cartoon Face Metadata Extension
- Cartoon Attributes for Machine Learning
Attributes
Original Data Source: Google Cartoon Face Metadata Extension
QUALITYLoading...