GitHub Icons Demo
Compare image hashing results between Dart and Python implementations using real GitHub icon datasets. Icons are clustered by their perceptual hash similarity.
Select Hashing Algorithm
Dart Implementation
Fast and simple hashing
Python Implementation
Fast and simple hashing
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
No corresponding cluster found
The Python implementation did not group these icons in the same way
How It Works
Collect Icons
Icons are collected from popular GitHub repositories like Eva Icons and Feather Icons.
Generate Hashes
Each icon is converted to PNG and processed through both Dart and Python implementations. Regular and z-transform variants are available.
Cluster Similar
Icons with identical hashes are grouped together, showing visually similar icons.
Z-transform variants (marked with orange border) apply histogram equalization preprocessing to normalize image brightness and contrast before hashing. This can improve clustering accuracy for images with varying lighting conditions.
Regular Algorithms
Process images directly with their original pixel values and lighting.
Z-Transform Variants
Apply histogram equalization first, then hash the normalized image.
This demo uses sample data showing how the icon clustering would work. The actual implementation processes thousands of real GitHub icons from repositories like Eva Icons and Feather Icons.