ellyuca / agv-project

AGV-Project for evolutionary adversarial attacks

AGV-Project

This is the official repository of the AGV-Project for crafting evolutionay adversarial attacks using Instagram inspired image filters.

AGV-emotions_attacks

Agv_emotions_attacks contains the code to attack an Emotion Recognizer. The Emotion Recognizer has been trained using the AffectNet dataset that can be downloaded from this link upon request.

Before running the code make sure to add the images for the adversarial attack in the following folder: datasets/images_dataset/emo_images/selected_balanced_images_renamed.

The weights of the trained model can be found in the folder: models/emo_weights.

1. Install dependencies.

pip install -r requirements_cpu.txt

To run the code on GPU, install this list instead:

pip install -r requirements_gpu.txt

2. Example of how to run the code:

python attacks_emotions/agv/agv_attack.py -l TEST.txt -o TEST.json -bs 1 -e 10 -pp 
"offsprings" -ps "direct" -po "ES"  -np 10 -el true -s pareto -df1 ssim  
-db "EMO-SUBSET-80-EMO-MOBILENET_V2-SINGLE"  -nf 3 -r true -lf TEST.out