Genetic Seam Carving

Genetic Seam Carving: A Genetic Algorithm Approach for Content-Aware Image Retargeting

Implementation of the genetic seam carving algorithm described in the references below. Genetic Seam Carving is an evolutionary algorithm for content-aware image resizing.

Requirements

Python 3

pip install -r requirements.txt

Usage

usage: main.py [-h] [--energy {sobel,scharr}]
               [--selection {roulette,tournament}]
               [--crossover {onepoint,twopoint,uniform}]
               [--mutation {uniform,shuffle,flipbit}] [--display]
               input target_shape target_shape output pop_size num_gens mut_pb

Genetic Seam Carving

positional arguments:
  input                 Input image
  target_shape          Target image shape in 'row col' format
  output                Output image
  pop_size              Population size
  num_gens              Number of generations
  mut_pb                Mutation probability

optional arguments:
  -h, --help            show this help message and exit
  --energy {sobel,scharr}
                        Energy map gradient
  --selection {roulette,tournament}
                        Selection operator
  --crossover {onepoint,twopoint,uniform}
                        Crossover operator
  --mutation {uniform,shuffle,flipbit}
                        Mutation operator
  --display             Display visualization

Examples

python3 main.py small_tower.jpg 242 286 smaller_tower.jpg 5 10 0.05

python3 main.py whale.jpg 340 408 whale_out.jpg 10 30 0.05 --crossover uniform

References

• Genetic Seam Carving: A Genetic Algorithm Approach for Content-Aware Image Retargeting
• An improved Genetic Algorithms-based Seam Carving method