Optimizing Camera Placement for Chicken Farm Monitoring

Prerequisites

• Python >= 3.12.4

Installing dependencies

For a CPU

pip install -r requirements.txt

For a GPU

pip install -r requirements-gpu.txt

Running the experiments

Performing all the experiments

python run_experiments.py all

Performing individual experiments

CMA-ES experiments

python run_experiments.py cma-es [OPTIONS]

Extra Options

• -e or --experiment The number of the experiment to be performed. Available experiments: [0 - 29].
• -g or --generations The total generations to execute the algorithm for. The population of each generation will be adjusted accordingly. Available generations: 16667, 1000, 2000.

If the option -g is given, all the experiments will be performed for that generation. If -e is given, the experiment will be performed for all generations. If both are given, only that specific experiment on the given generation will be performed.

Examples

• All experiments for 1000 generations.

python run_experiments.py cma-es -g 1000

• The 2nd experiment will be performed for all generations

python run_experiments.py cma-es -e 2

• The 21st experiment will be performed for 16667 generations. One experiment will be performed.

python run_experiments.py cma-es -g 16667 -e 21

Quality Diversity experiments

python run_experiments.py qd ALGORITHM BEHAVIOUR_DESCRIPTOR [OPTIONS]

• Available options for ALGORITHM:
  • isolinedd
  • cma-me
• Available options for BEHAVIOUR_DESCRIPTOR:
  • used-beams
  • spread-xy

Extra Options

• -e or --experiment The number of the experiment to be performed. Available experiments: [0 - 29].
• -g or --grid_size The grid size of the behaviour space. Available size: 1000, 10000. It can only be used with spread-xy.

If -e is given, the experiment will be performed for all grid sizes. If the option -g is given, all the experiments will be performed for that grid size. If both are given, only that specific experiment on the given generation will be performed.

Examples

• All experiments using CMA-ME, spread-xy, and 1000 grid size.

python run_experiments.py qd cma-me spread-xy -g 1000

• The 28th experiment using IsoLineDD, spread-xy, and 10000 grid size.

python run_experiments.py qd isolinedd spread-xy -e 28 -g 10000

• The 14th experiment using CMA-ME, and used-beams.

python run_experiments.py qd cma-me used-beams -e 14

Plotting results

• To only display the final solutions (the state of the algorithms after their final generation).

python plot.py solutions PATHS... [OPTIONS]

• To display the performance of the experiments over time.

python plot.py overtime PATHS... [OPTIONS]

• To create a CCDF (Complementary cumulative distribution function) of the experiments.

python plot.py ccdf PATHS... [OPTIONS]

Note: for over time experiments, CMA-ES experiments cannot be combined with QD experiments, and vice-versa.

Extra Options

• -f or --filename Optional filename for the plot.