This project implements a genetic algorithm to reconstruct target images by evolving populations of abstract shapes. Over successive generations, the algorithm optimizes the visual resemblance between a rendered image and a given target, showcasing the creative power of evolutionary computation.
Evaluates the similarity between the target and candidate image using pixel-wise comparison.
- Selection: Chooses the fittest individuals for reproduction.
- Crossover: Combines traits from parents to generate offspring.
- Mutation: Introduces random variations to maintain diversity.
Observe the approximation progress live as the population evolves.
Easily switch between different images to test algorithm scalability and generalization.
Python 3.x
Required libraries:
randomPILnumpymatplotlibcv2mathtabulate
EvoPaint demonstrates the power of nature-inspired optimization.
It’s a visual and computational experiment in how simple rules and randomness can produce sophisticated image approximations.
This project was developed as part of the Artificial Intelligence coursework at the University of Tehran.
Instructor: Dr. YaghoobZadeh