Anonymizer

An anonymizer for pravicy protection of human faces and vehicle plates. This is a customized fork originally from archived project of understand-ai. The original code is developed with tensorflow-1.11 which is a fairly old version with poor GPU support in recent GPUs. With some tricks, the ONNX-RUNTIME provides a work around to running on fairly new GPUs (i.e. CUDA 11.*). Guidelines for ONNX-RUNTIME version is here.

Installation

Preparation

• Clone the Repo

  git clone https://github.com/fusionportable/Anonymizer.git

• Download the pre-trained weights

  # two .pb files
  weights_plate_v1.0.0.pb
  weights_face_v1.0.0.pb

• Build Docker Container

  # build from local
  docker build -t ros-noetic-miniconda:latest .
  # if build failed pull from docker hub
  docker pull 11710414/ros-noetic-miniconda:latest

Install Dependencies

conda create --name anonymizer python=3.6 -y
conda activate anonymizer
cd /workspace/anonymizer
pip install --upgrade pip
pip install -r requirements.txt

TODO:

• A docker container with ROS1 and CUDA11.6 compatibility.
• Add instructions on easy run
• Add figures for illustration
• Add scripts directly process the rosbags.
• Refactor the anonymizer in the same branch that is compatiable to plug in detector and obfuscators.