sbnsl / FLM Public

forked from alldbi/FLM

Notifications You must be signed in to change notification settings
Fork 0
Star 2

Tensorflow implementation of Fast Geometrically-Perturbed Adversarial Faces

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 23 Commits
.idea		.idea
src		src
utils		utils
LICENSE.md		LICENSE.md
README.md		README.md
__init__.py		__init__.py
main.py		main.py

Repository files navigation

FLM

Tensorflow implementation of "fast landmark manipulation method" (FLM) and "grouped fast landmark manipulation method" (GFLM) for generating adversarial faces, from our paper: [Fast Geometrically-Perturbed Adversarial Faces](address will be added).

Setup

Prerequisites

Tensorflow 1.4.1
Dlib
CV2

Pretrained models

Download pretrained "Inception ResNet v1" model, trained on the "CASIA-WebFace" dataset provided by facenet.
Download the file and extract it to get the "shape_predictor_68_face_landmarks.dat" pretrained model for DLib landmark detector.

Getting Started

# clone this repo
git clone https://github.com/alldbi/FLM.git
cd FLM

# Generating adversarial faces by Grouped FLM:
python main.py \
  --method GFLM \
  --output_dir facades_train \
  --max_epochs 200 \
  --input_dir facades/train \
  --which_direction BtoA